Guide to Linear Electronics: Power supply systems

Home | Glossary | Books | Links/Resources
EMC Testing | Environmental Testing | Vibration Testing


Active components operate by transforming energy supplied by a power source into some other desired form. Because of this, the way they operate must always be dependent, to some extent, on the quality and purity of the power supply by which they are energized, although one of the aspects of circuit design can be to minimize, where possible, the influence of the power supply on the behavior of the circuit.

Most electronic circuitry will be designed to be powered by some form of DC supply arrangement, although it’s possible to design some kinds of equipment which will work satisfactorily with a raw AC voltage supply rail, and this may be done where low cost is the major design objective. Of the DC sources, the commonest forms are 'primary' (expendable), or 'secondary' (rechargeable), batteries, AC mains powered DC supplies, and solar cells.

In general, the choice of supply type will depend on the total energy requirement, and whether the equipment needs to be fully portable. Solar cell power supplies are restricted to use in medium to high light level environments, and with systems, such as electronic calculators, which require very little power, though output efficiency improvements in such cells are changing this situation.

Batteries (groups of electro-chemical cells, connected in parallel or series) allow a higher output power, but are either expensive to replace or require recharging at intervals. Their use also carries with it the possibility of damage to the equipment with which they are used through leakage of corrosive electrolytes, though improved battery design has led to 'leak free' systems.

AC (mains) supply line operated power supplies will allow an almost unlimited range of available output powers, but they may be heavy, and they will usually restrict the portability of the equipment. They also lead to the possibility of mains hum intrusion into the electrical output of the circuit, either because of the presence of residues of mains frequency voltage ripple on the DC output rails, or by interaction between the alternating magnetic field associated with the mains transformer in the power supply and the internal electrical circuit connections in the equipment with which it’s used.

Battery power supplies:

Depending on the relative costs and service requirements the choice will lie between primary cells -- those types which are used until their output voltage falls below the minimum satisfactory level, but must then be removed and discarded -- and secondary cells, which can be repeatedly recharged.

Primary cells:

Of the primary cells, the choice is between cost and performance -- in terms of their output power to weight ratio, their constancy of voltage output as a function of time, electrical load, and ambient temperature, their output voltage, and their resistance to leakage of electrolyte. The major contemporary cell types are listed below.


A wide range of 'voltaic' cells was devised, during the early part of the 19th century, based on the combination of various pairs of metals with some acid or alkaline electrolyte, following Alessandro Volta's discovery in 1799 that zinc and silver metal discs, separated by a layer of moist cloth, would generate an externally measurable electrical potential. Most of these cells, such as those due to Daniell, Grove and Bunsen, remained little more than laboratory curiosities, but the system devised, in 1866, by Georges Leclanche which used an electrode pair of carbon (+ve), and zinc (-ve), in an ammonium chloride electrolyte, proved exceedingly successful, and has, with some improvements, been adopted as the basis for the vast majority of inexpensive 'dry' batteries.

In its simplest form, the construction adopted is as shown in FIG1, in which a U-shaped thin-walled tube of zinc, is lined on its inside with blotting paper, and is used to hold either an ammonium chloride or a zinc chloride electrolyte, or a mixture of these, with some liquid absorbent or gelling agent, to reduce the likelihood of leakage. Also contained inside the tube is a cloth bag holding the positive electrode, which consists of a carbon rod, surrounded by a mixture of graphite, to help conductivity, and manganese dioxide, which operates both as a 'depolarizing' agent -- to prevent the evolution of gaseous hydrogen at this electrode -- and as a source of oxygen to activate the cell. The cell is completed by crimping a tin plated brass cap on the top of the central carbon rod electrode, sealing the gap between the rod and the outside tube at the top of the cell with pitch, and wrapping a paper label around the outside of the zinc tube. Such cells are cheap to make, have a storage life of up to a year, require little exotic manufacturing technology, and have an adequate output power to weight ratio.

Cylindrical zinc anode (-ve) -- Cardboard sleeve

Mixture of graphite and ???2 impregnated with ammonium chloride solution

Cloth bag

Carbon rod (cathode)

FIG1 Simple dry battery (Leclanche cell)

The initial output voltage for a new cell will be in the range 1.55-1.6V, which will gradually fall, during use, to 1.1-1.2V, by which time its service life must be considered to be at an end, since any further discharge beyond this point will cause the output voltage to collapse rapidly to a near-zero level. In more recent cell technology, zinc chloride has largely come to replace the ammonium chloride electrolyte, because it gives a somewhat greater cell capacity: a simple modification which allows the makers to charge a higher price. During the discharge process, the manganese dioxide is reduced to manganese hydroxide, and the metallic zinc electrode is either oxidized to a basic zinc manganate, or combines with the ammonium chloride to form a diammine chloride. If a zinc chloride electrolyte is used, the zinc is mainly converted into zinc oxychloride. In either case the cell becomes discharged when most of the accessible metallic zinc is consumed, or when the bulk of the manganese dioxide is reduced to an unreactive state. Similar processes occur in all of the other cells using a zinc anode, except that those using a potassium hydroxide electrolyte will ultimately convert the metallic zinc into zinc hydroxide or oxide.

The major problem with the simple Leclanche type of cell is that since zinc is consumed during the discharge process, the outer zinc tube will, in due course, become eroded, and will perforate. This is hazardous if the cell is allowed to remain in situ in a discharged condition, since a perforated outer shell will allow the corrosive electrolyte to creep into other parts of the equipment in which it’s used, and this can be exceedingly destructive if not caught in time.

A more modern type of construction, shown in FIG2, uses an extruded plastic tube, crimped over tin plated iron top and bottom caps, both to contain the cell, and to form a short-term barrier to the leakage of the electrolyte if the zinc tube should perforate. With improved purity in both the metal used for the zinc tube and the electrolyte chemicals these cells have a much greater storage life -- up to 3-4 years at temperatures up to 20°C. The useful range of working temperatures for this kind of cell is effectively between 5° and 30°C, since with all of these cells the life expectancy and retained capacity are lowered above 30°C, while both the ampere/hour capacity and the maximum output current obtainable from the cell are greatly reduced below 0°C.

Metal top cap

Bitumen seal

Zinc tube

Insulating sleeve

Graphite and manganese dioxide filling. Impregnated with zinc chloride solution

Bottom washer and separator

Metal bottom cap

FIG2 Improved form of dry cell. The zinc chloride cell

Another common style of this type of battery is the layer cell type, shown in FIG3, in which a stack of cells -- usually of four or six -- made in pancake form, and individually contained in a plastics (i.e. an insulating synthetic polymer such as polypropylene or rigid PVC) tube, are mounted within a plastic lined sheet metal jacket, and internally connected to give an output voltage of 6V or 9V. This type of battery was very popular as a source of power for portable radios, but has been rather overtaken by circuit design improvements which allow operation from one or two of the less expensive cylindrical form cells.

Outer case; Individual battery packs

FIG3 Layer cell battery

Alkaline manganese cells

While still a zinc/carbon/manganese dioxide system, with a nominal output voltage of 1.55V, this style of cell, shown in cross-section in FIG4, gives a substantial step forward in performance in that it gives almost double the storage capacity and shelf life expectancy of the normal zinc/carbon Leclanche cell.

Steel cap and +ve connector Anpde-naiV Negative_ contact C2S-J Cathode pellets

- (Mn02 + graphite press molded)

Shrink fit sleeve

Sealing grommet

FIG4 Alkaline -- manganese cell

To the user, the most conspicuous visual difference is that the cell construction is reversed, with the positive terminal being a protruding dimple formed on the outer metal case, while the negative contact is a cup-shaped disk, insulated from the steel outer case by a plastic sealing washer.

In its construction, the extruded zinc container of the simple Leclanche cell is replaced by a thin sheet steel tube inside which, and in intimate contact with it, is a stack of cylindrical cathode 'pellets', formed under pressure from a mixture of graphite and manganese dioxide. An inner sleeve of porous insulating material separates this from the anode electrode, which is a filling of zinc powder, formed around the zinc current collecting 'nail', and made into a paste with the potassium hydroxide solution electrolyte.

From the point of view of the user, a major advantage, which offsets its greater purchase price when used with expensive electronic equipment, is that the outer steel case is resistant to the electrolyte, and the cell is therefore leak proof under all normal conditions of use. Also, because of the greater efficiency and conductivity of the potassium hydroxide electrolyte, the internal resistance of the cell is lower, which allows greater peak output currents to be drawn for brief periods. However, apart from its greater ampere/hour capacity for a given cell size, the characteristics of the alkaline manganese cell are very similar to that of the Leclanche.

Button cells

Mercuric oxide systems

Both the zinc-carbon systems and the nickel-cadmium rechargeable cells, examined later in this section, tend to be relatively bulky. This is not a particular disadvantage in the case of hand torches, or most of the cassette recorders, cordless telephones or portable radios with which they will normally be used.

However, the advent of miniaturized hearing aid systems, electronic wrist watches, credit card sized pocket calculators, and electronic exposure and shutter control systems in small cameras, created a demand for cells of very much smaller dimensions, and with a much more constant voltage output. Of these, one of the earliest to be exploited commercially was the mercuric oxide-zinc cell, of which the basic system had been invented in 1886 by Aron, and further developed in the 1930s by Samuel Ruben, although Ruben's interest was in the relative constancy of the output voltage of this kind of cell, rather than its capacity for miniaturization.

The method of construction of a contemporary cell of this type is shown in FIG5, and consists of a cathode of mercuric oxide, with some graphite to increase its conductivity, compressed into a small flat pellet, pressed into the base of a tin or cadmium plated steel can, and separated from a high purity zinc-mercury amalgam anode by an absorbent pad containing the potassium hydroxide electrolyte. This kind of cell has an output voltage of 1.3-133V, depending on output current, which will remain constant up to the point at which it’s almost completely discharged. This gives the sometimes disconcerting characteristic, for example in electronic wrist watches, that when they do stop, nothing short of replacing the cell will cause them to re-start.

Sealing grommet Steel cap to cell (-ve contact)

Cell can (+ve) amalgam anode; Absorbent pad to hold electrolyte "Ion permeable-separator Mercuric oxide cathode

FIG5 Mercuric oxide cell Silver oxide-zinc cells

This is very similar in construction and application to the mercury cell, and was also derived from a design originally patented in the mid-1800s (in this case by Clarke) developed in the early years of the 1939-45 war by Andre, as a miniature voltage source for military applications. In this cell, the cathode is a compacted pellet of silver oxide and graphite -- added to increase conductivity. Its output voltage is 1.55V, and, like the mercury button cell it has a very flat output voltage vs. discharge characteristic.

Zinc-air cells

By using atmospheric oxygen, adsorbed onto a catalytic carbon composite layer, as the cathode, more of the internal cell space can be used for the consumable zinc amalgam anode, leading to a greater power capacity for a given volume. The basic construction, shown in FIG6, is similar to that of the mercury cell, apart from the fact that the bottom of the cell case is perforated to provide access for air. During storage these perforations are covered with a layer of impermeable tape, which must be removed to activate the cell. However, once the cell has been activated it will continue to discharge slowly, so long as these access holes are uncovered -- a characteristic which makes such cells more suited for applications requiring continuous rather than intermittent use. The output voltage of this type of cell is 1.2-1.4V depending on load current, and it has a discharge voltage characteristic which is similar to that of the mercury cell.

Zinc amalgam anode impregnated with electrolyte; Sealing grommet; Separator L Cell can (+ve)

FIG6 Zinc-air cell

Lithium cells:

The search for very high power-to-weight ratios in primary cells has led to the exploration of a number of exotic electro-chemical combinations, of which the most successful have been those based on metallic lithium in combination with a wide range of oxide, sulphide or fluoride cathodes. Because lithium will react vigorously with water, it’s necessary to find some non-aqueous solvent for the electrolyte, such as dimethoxyethane.

A commercially available cell (the Ever Ready type 2016 button cell) employs a manganese dioxide cathode and a lithium perchlorate electrolyte to give a 3.2V output voltage, and good storage and discharge voltage characteristics. Similar cells, such as the Crompton 'Eternacell', are available which employ a thionyl chloride electrolyte.

All of these Lithium cells are characterized by a high energy to weight ratio -- up to three times better than the alkaline manganese cell -- a wide operating temperature range (typically -50°C to +60°C), and an exceedingly long storage life, which can be in excess of ten years. This feature is exceedingly useful where a permanent voltage source is required in applications where only a very low output current demand is likely, such as, for example, as a back-up voltage source for 'volatile' computer memory banks, to protect against data loss during a power failure.

It’s essential to avoid reverse (charging) current flow through these cells, an action which can lead to the cell exploding, so the circuit shown in FIG7 is normally used in back-up applications to prevent this happening.

Computer memory; Lithium back-up cell

FIG7 Lithium cell computer memory back up circuit

Secondary cells:

Those types of cell which can be recharged to restore their initial output voltage and stored energy capacity are known as secondary cells, to distinguish them from the use and throw away types, known as primary cells.

Although it’s possible to recharge standard Leclanche type dry cells, provided that they are not too fully discharged (see Wireless World, August 1981, p. 70, and Feb. 1982 p. 46), the results are not as satisfactory as in the case of cell systems specifically designed for this purpose.

Of the secondary cells which have achieved more general popularity, the major types are the lead-acid, the nickel-iron, or NiFe, and the nickel-cadmium, or NiCad cells. Of these, the lead-acid types are by far the most common, in that they are the standard source of power for starting and lighting motor vehicles.

The specification of secondary cells generally includes both their nominal output voltage and their discharge capacity, in milliampere or ampere hours. In all cells, this capacity is dependent on the effective surface area of the reactive materials on the plates, so means are taken to make this as large as practicable, so far as this is compatible with mechanical robustness and impact resistance. The discharge capacity rating is not usually quoted for primary cells, mainly because it depends on so many factors, such as operating temperature and discharge rate, but, as a general rule, the ampere hour capacity of a D size sealed lead-acid cell would be of the order of 2-2.5Ah, as compared with 4Ah for a NiCad cell, 7Ah for a zinc chloride electrolyte Leclanche type, and 15Ah for an alkaline manganese cell, when discharged intermittently at a 30mA output current.

Lead-acid systems:

In the lead-acid type of cell, the electro-voltaic couple is between lead dioxide (PbO2) and metallic lead, using a dilute sulphuric acid electrolyte, of which the specific gravity, SG, when fully charged = 1.26. This gives an output voltage -- when fully charged -- of about 2.25V/cell, decreasing to about 1.9V/cell when approaching the discharged state.

In a typical open cell system the positive plate electrode consists typically of a hollow lead matrix, of rectangular or honeycomb type cavities filled with lead peroxide, with a negative plate held out of direct electrical contact with the positive one by means of a porous 'separator', having a pocketed or spongy surface. The construction of such a cell is shown, schematically, in FIG8. The chemical reaction which occurs during discharge is, theoretically, that the lead dioxide is reduced to metallic lead releasing oxygen, while the lead (negative) electrode is oxidized to lead sulphate, releasing hydrogen. This action consumes part of the sulphuric acid, and releases water by electrolytic recombination, so the specific gravity (SG) of the electrolyte falls during discharge, to a nominal SG when fully discharged, of 1.1. If the cell is allowed to stand for long in a discharged condition, both of the plates may become covered in lead sulphate, and the electrolyte thus further depleted of acid has an increasingly high resistance, which restricts the possible current flow, and makes recharging difficult.

This is described as the battery being 'sulphated'.

Inter cell connections; Acid proof container; Sulphuric acid (20%); Metallic lead (-ve) electrode; Porous cell separator; Lead peroxide (+ve) electrode

FIG8 Basic lead-acid cell

In normal use, the ampere hour capacity of lead acid cells is gradually reduced by the loss of the lead dioxide coating on the positive plate. However, the normal failure mechanism for lead-acid cells is that this lead dioxide, when shed by the positive plate, and then partially reduced to metallic lead, forms a conductive bridge between the plates which permits the battery to discharge even under no-load conditions. In the case of automobile batteries, one might suspect that the design of the battery is chosen to permit, or even encourage, this type of failure mechanism, to prevent the cells from lasting too long! Ideally the level of current used to recharge a lead acid battery should be between C/5 and C/20, where C is the ampere hour rating of the cell. These limits would be referred to, respectively, as 'five hour' or 'twenty hour' charging rates. In emergencies, higher currents can be used, up to 2C, but both high charging and high discharging rates tend to accelerate the loss of lead dioxide from the positive plates. Since both the charge and discharge mechanisms are less than completely efficient, 'outgassing' (the electrolytic evolution of gaseous hydrogen and oxygen) occurs during both the charge and discharge cycles, especially at high current levels. This is unimportant in Open-cell' car battery systems, where provision is made to allow such gases to vent to the atmosphere, with the lost water being replaced as necessary. At high charge/discharge rates the electrolyte may be lost as spray, and simple topping up with distilled water will gradually cause the electrolyte to become increasingly dilute.

In 'sealed' lead-acid cells, the electrolyte will either be held in some absorbent filling material, or as a gel, in order to lessen the possibility of leakage. The type of construction adopted will also be chosen to promote electrolytic recombination. It’s probable also that the manufacturers will recommend a maximum charge/discharge current to help prevent too rapid a rate of evolution of gases.

Lead-acid cells have the advantage of a relatively high output voltage and, when new, a very low internal source resistance, but worries about possible electrolyte leakage have tended to make designers choose NiCad types, whose immunity form this type of problem has been more fully established, even though improved sealed lead-acid systems have now been shown to be exceedingly reliable and trouble-free.

NiFe and NiCad systems

The exploration of systems based on an alkaline hydroxide electrolyte, by Waldemar Jugner in Sweden, and Thomas Edison in the USA, led to the use of both nickel hydroxide/iron and nickel hydroxide/cadmium couples in open-cell batteries, with Edison's NiFe cells being employed first in the early 1920s. These were relatively inexpensive and exceedingly robust electrically -- indeed they had the reputation of being virtually indestructible, but had poor charge efficiency, which led to rapid loss of electrolyte through outgassing, and a relatively low energy density -- only about 25% of the theoretical utilization of active material being practicable at that time. In recent open-cell NiFe types better energy densities of up to 40 50Wh/Kg have been obtained. The output voltage discharge curve of such cells falls from 1.3 to 1.15V at a C/6 discharge rate. A very popular application for such cells was in the 120V 'Milnes' radio HT unit, popular in the 1930-40s as a source of anode voltage in battery operated radio sets. The arrangement of the cells in this unit-individually contained in small glass jars in a wooden case -- was such that they could be switched into parallel groups for recharging from a standard 6V car battery, or back again into series to provide the 120V output.

The use of this type of system in sealed cells was not easily possible, and this factor, together with the improved NiCad energy capacity, led to the greater popularity of the nickel/cadmium types, which can be designed so that there is very little evolution of gas in normal use. The construction of these cells is shown, schematically, in FIG9. In this, both the negative (cadmium) and the positive (nickel) plates are formed from a microporous mass of powdered metal, with the nickel electrode being largely oxidized during manufacture to basic nickel oxide (NiOOH). The chemical reaction which occurs during discharge is that this oxide is reduced to the hydroxide (Ni(OH)2), while the released oxygen diffusing through the cell oxidizes the cadmium negative plate to its hydroxide, (Cd(OH)2) --a process which is fully reversible. Although water, from the potassium and lithium hydroxide solution which forms the electrolyte, is involved in this process, it’s not consumed in normal use, so, in sealed cell manufacture only the amount judged to be theoretically necessary is provided.

Positive connector Nylon sealing grommet Sintered +ve electrode Separator Nickel plated steel can connected to -ve electrode

FIG9 Cylindrical form Nickel-Cadmium cell

If charging is continued beyond the fully charged condition surplus cadmium on the negative plate is able to convert the oxygen evolved on the nickel plate back to water, only heat being evolved. The warm-up of the cells can be used as an indication that they are fully charged.

The major failure mechanism in NiCad cells is the loss, by electrolysis, of the available water in the system as a result of inadvertent reverse charging -- a process which can happen, all too easily, if current is drawn from a partially discharged battery which contains one cell which has become more completely discharged than the others, but can also occur if the cell is excessively over-charged.

A further mechanism to which more gently used NiCad cells seem particularly prone is the growth of fine threads of conducting metallic cadmium -- known as 'dendrites' because of their tree-like structure --through the porous separator between the plates, which not only causes the cell to discharge, but may also prevent its being charged again because of the internal shortcircuit which is present.

These failure mechanisms were examined in detail by Cooper (Wireless World, May 1985, pp. 61-63, June 1985 pp. 60-63, and July 1985, pp. 32-36.), and the means of restoring internally short-circuited cells to working order, by subjecting them to bursts of charging current large enough to fuse the cadmium dendrites is examined by Johnson (Wireless World, February 1977, pp. 47-48). On the credit side, such cells are relatively light and mechanically robust, largely trouble free in normal use, and are capable of very high output currents over brief periods, if needed.

Indeed, a very successful range of portable soldering irons has been marketed, powered by NiCad cells housed in the handle, which supply a current to the low resistance heater element which is high enough to bring the bit up to operating temperature within a few seconds.

Various charging systems have been suggested for NiCad cells, but the most straightforward is to re charge them separately at a fixed voltage of 1.42V/cell, when damage due to overcharging is not possible. Where fixed voltage charging is not feasible, current charging rates in the range C/5 to C/10 are usually recommended -- preferably combined with a cell temperature rise cut-out mechanism.

The output voltage discharge curve is very flat, falling from 1.22V to 1.19V at the threshold of discharge, at a discharge current of C/5. Immunity to self discharge is reasonably good in modem cell designs, though less satisfactory in sintered plate systems.

Other rechargeable cells have been devised, based on, among other possibilities silver-zinc (1.5V), silver-cadmium (1.1V), and Nickel-Zinc (1.8V) types, but these remain relative rarities.

I have illustrated in FIG10 the relative output voltage discharge curves for the more common primary and secondary cell types, scaled to represent the performance which would be given by cells having an equivalent physical size, and discharged at the same C/JC current flow.

I would like to express my gratitude to the Ever Ready and Duracell battery companies for their generous provision of technical information on this subject.

FIG10 Voltage output and energy capacity for common cell types

FIG11 Simple transformer-rectifier power supplies

Mains operated power supplies

Simple transformer-rectifier systems

The need to provide a stable, reasonably noise and ripple free DC output from a 50 or 60Hz AC power supply line is capable of being met by a wide range of circuit layouts, but, in general, the designer must first decide how good a quality of output DC is needed by the system to be powered, and to what extent a more expensive or complex power supply layout would be justified by circumstances.

In its simplest form, a DC output can be provided by any of the transformer/rectifier/capacitor layouts shown in FIG11. A point which must always be borne in mind, in any rectifier/reservoir capacitor layout, is that, on light load, the output voltage will be 1.414 times the RMS AC output voltage provided by the transformer secondary winding. This means that the energy output for which the transformer is rated, in volt-amperes, usually abbreviated to VA, must be down rated by a factor of, at least, 1.414 to allow for the fact that the output current from the power supply is being drawn, effectively, at a higher voltage than the secondary RMS output value. This means that, in the case of the simple circuit layout of FIG11a, if the secondary winding is rated at, say, 1A, the maxi mum output current which can be drawn from the supply will be 0.707A, even though, under load, the resistive and core losses in the transformer, and the inevitable forward (conducting), voltage drop across the rectifier diode, will mean that the voltage actually developed across the reservoir capacitor, C1 will actually be a good bit less than 1.414 Vout. Also, in practice, the maximum output current will be reduced even further because the current drawn from the secondary winding of the transformer will take the form of short duration high-current pulses, as shown in FIG12. It must be remembered that the core and winding resistance losses are proportional to T2 rather than I, so the high output current pulses drawn from the secondary winding will cause a disproportionately high transformer dissipation.

FIG12 Voltage and current waveforms --half wave rectification.

The allowance which must be made for this effect cannot be precisely specified because it depends on both the transformer core characteristics and the size of the reservoir capacitor employed -- the larger the capacitance of this the shorter in duration and the larger in magnitude the repetitive charging current pulses will be. A working rule of thumb is that the DC output current (f_dc) should not exceed 0.65 /ac. This leads to the need for a design compromise, in that, if the DC output is to be drawn directly from the reservoir capacitor, as is increasingly the case with modern transistor and IC power supplies, the desire to keep the output ripple level low would urge the use of a large value reservoir capacitor, but, because this will cause the current drawn from the mains transformer secondary to be reduced to progressively shorter reservoir capacitor charging pulses, this worsens the transformer efficiency and heat dissipation. However, the inevitable secondary winding resistance, coupled with core saturation effects, means that, in practice, there is an upper limit of reservoir capacitor size, beyond which no further benefit will be obtained.

The 'choke-input' filter system shown in FIG13 makes much better use of the potential power output from the transformer, but the DC output voltage developed across the reservoir capacitor, C1 will al ways be somewhat less than the transformer RMS secondary voltage. Also, because of the need for a gapped core 'swinging' choke -- a somewhat rare type of component, which will add to the bulk and component cost of the circuit -- this type of power supply system is rarely found in low to medium output power designs.

FIG13 Choke-input filter system.

Where just a single line or a pair of +/- supply lines are needed, the most common types of circuit are those shown in FIGs 11b, c, d and e, where 'full-wave' rectification is employed to double the frequency of the charging pulses fed to the reservoir capacitor, and nearly halve the magnitude of the residual sawtooth supply line ripple. A comparison which is illustrated in FIG14.

FIG14 Voltage and current waveforms. Full wave rectification

Even with full-wave rectification, and a large value reservoir capacitor, there will always be a significant amount of (100/120Hz) ripple on such a DC supply output, and it’s inevitable that the mean output voltage will fall and the ripple content will worsen as the output current demand is increased. However, such simple systems are very widely used and, with care in the design of the circuitry to be used with them, are often quite adequate for their purpose.

In valve operated electronic systems, where relatively high DC supply line voltages were used, and the DC current demand was therefore proportionately lower, it was conventional practice to insert a 'smoothing choke' (L1), between the reservoir capacitor (C1), and the output 'smoothing capacitor' (C2), as shown in FIGlid. This converts L1/C2 into a-12dB/octave LC type low-pass filter and substantially reduces the amount of residual sawtooth HT ripple present in the output circuit: a ripple voltage which is caused by the way in which the rectifier/reservoir capacitor charging mechanism acts.

However, with high current, low voltage, power supplies the additional DC resistance introduced into the output path by a series choke of adequate inductance to be useful would be undesirable, and the use of such series connected chokes has become a rarity; particularly since output voltage smoothing can be provided in a much more compact and economical way, if needed, by the use of a voltage regulator circuit.

Voltage regulator systems:

It’s often necessary to supply the operating circuitry with a substantially constant, and ripple free, DC voltage -- a requirement which cannot be met by any simple mains transformer/rectifier layout. Such improved supply stability can be provided by the use of some kind of stabilizer or regulator circuit, and these can be divided into 'shunt' and 'series' systems.

Shunt regulator layouts:

The simplest example of this type of system is the zener or avalanche diode circuit shown in FIG15. (Note. Although true zener diodes are only those with a turn-over voltage of about 5V or below, the term zener has come to be used as a generic term for all reverse breakdown regulator devices.) Such a layout is called a shunt regulator because the voltage sensitive component/s are connected so that they will draw current in parallel with the load, and will increase their current flow if the load current is reduced or the output voltage tends to increase.

In the circuit of FIG15, the input supply voltage, V^, the value of the series resistance, R19 and the power handling capacity of the zener diode, D1? are chosen so that the required output voltage will be provided, without exceeding the permitted dissipation of the diode, at zero output current demand, but so there will always be an adequate current flow through the diode at the maximum likely load current. The output source resistance is determined by the slope of the diode voltage, Vd, when plotted against diode current, 7d, as shown in FIG16. For example, if the diode characteristics are such that an increase in current of 100mA causes an increase in the voltage developed across the diode of 0.1V, the effective source resistance will be one ohm, the attenuation of the input voltage ripple, for the circuit values quoted in FIG15, will be 50:1, and the required dissipation for D1 will be 0.6W. Unfortunately, for reasons examined in Section 4, both zener and avalanche diodes generate a small amount of wideband noise, so it’s probably worthwhile to connect a capacitor, C1? in parallel with the diode output to reduce somewhat this noise contribution.

FIG15 Simple zener diode shunt regulator system

Diode current (mA)

FIG16 Typical zener diode regulation characteristics

For low voltages a simple forward biased silicon junction diode is usable as a reference source, and will give an output voltage between 0.55V and 0.65V, depending on diode current, with a very low electrical noise contribution. Several such diodes could be connected in series, as shown in FIG17, to give, for example, a 2.4V DC source.

FIG17 Forward biased diode chain

The output characteristics of the simple zener diode circuit of FIG15 can be substantially improved by connecting the diode in the base circuit of an amplifying transistor, as shown in FIG18. This also allows a low-power diode to be used, which is useful since medium to high power transistors are usually cheaper than zener diodes having the same dissipation ratings. Also, since voltage regulator diodes, with turn-over voltages above some 5 V, have a positive temperature coefficient of output voltage -

depending somewhat on diode current for a given junction area, as shown in Fig 15.19, -- while normal silicon junction diodes, such as the base-emitter junction of a transistor, have a negative temperature coefficient, connecting these in series, as in FIG18, can partially compensate for the positive temperature coefficient of the diode.

Stabilized DC output

FIG18 Improved zener diode shunt regulator

Diode voltage (V) FIG19 Temperature coefficient of zener diode regulator Another simple shunt regulator layout is the so called 'amplified diode' circuit shown in FIG20. In this the transistor conducts if the voltage developed across R3 exceeds some 0.55-0.6V, giving an output voltage of Vbe (R2+^3)^3, which is adjustable, within limits, by the choice of values of R2 ovR3. This type of layout is commonly used as a method of stabilizing the forward bias on the output transistor pair in class AB audio amplifiers, where Q1 will often be mounted in thermal contact with the output transistor heat sink, in order to provide a measure of operating temperature compensation.


DC output

FIG20 The amplified diode circuit

More elaborate shunt regulator systems, of the type shown schematically in FIG21 -- in which the output voltage is compared with a precision voltage reference source, and the difference voltage is amplified and used to control a shunt transistor load -- can be used to provide improved regulation characteristics, but this layout is only likely to be used where the difference between the available input supply volt age and the required output voltage is so small that a series regulator circuit would be impracticable.

Series voltage regulator systems:

FIG21 High stability shunt voltage regulator circuit.

This type of circuit layout is one in which a series transistor, or equivalent device, is interposed between the input voltage source and the output point, so that, if the output voltage tends to rise above, or fall below the target figure, then the current flow through the series (pass) transistor can be adjusted by some external circuit so that the required output voltage level is maintained.

The simplest feasible circuit for this purpose is one in which the base of the pass transistor is fed from a fixed voltage source, such as from a simple zener diode voltage regulator, as shown in FIG22. If the output voltage falls, the base-emitter voltage applied to Ch will increase, and so will the current through Qj. If the output voltage tends to increase, then both the forward bias on Q1? and the current flow through Q1 will decrease.

FIG22 stabilizer

Simple zener regulated series

There are three main snags with this circuit, of which the first is that R1 must be able to pass a sufficient current to meet the base current demand of Q1 at the required maximum output current demand, plus what ever minimum current is required to cause the zener diode to operate. Secondly, the value of Rx cannot be made too low, or otherwise the desired attenuation of the input noise and ripple will be greatly lessened, and This means that the input voltage must exceed the required output voltage by an adequate amount.

Thirdly, the voltage across the zener diode will drop as more current is withdrawn by the base of the pass transistor, and the base-emitter voltage drop of Q1 as the output current demand is increased, will also become larger. Both of these effects operate in the same sense to reduce the output voltage provided by the circuit, and this gives the circuit a relatively high output (source) impedance.

A more conventional series regulator circuit will normally use the circuit layout shown in FIG23, in which the base current of the pass transistor is derived from the output of a voltage amplifier, whose input circuit is connected so that it’s able to sense the difference between its input voltage, -- usually related to the regulator output voltage -- and some stable and noise-free voltage reference source. If the output volt age from the regulator should vary, the amplifier will operate to oppose this variation. By the use of an error amplifier having sufficient gain a high degree of out put voltage stabilization, and a very low output impedance, typically less than 0.02 ohms, at frequencies up to a few kHz, can be obtained.

FIG23 Schematic layout of conventional series stabilization circuit.

The major drawback with this type of layout, in which the pass transistor is connected as an emitter follower, is that enough voltage difference must exist between the available input potential and the required output voltage for the amplifier to be able to drive the pass transistor adequately. The minimum input-output voltage requirement is commonly termed the 'drop out' voltage, and, with a well designed circuit can be as low as 2-3V.

The problem of the need to allow enough voltage drop in the circuit to provide forward bias for the transistor can be lessened by using the transistor in an inverted mode, in which the output current is drawn from its collector, so that the forward operating bias is derived from the potential difference between the input supply lines, as shown in FIG24. By comparison with the circuit design of FIG23, the intrinsic output impedance of this layout is higher because the load is fed from the collector of the pass transistor, which is a high impedance point. By the use of a high system gain, however, output impedances below 0.05 ohms can still be obtained.


Over-current protection systems:

A problem which is inherent in any series regulator system is that, in the event of an output short circuit, a very heavy instantaneous current could flow from the power supply reservoir capacitor through the pass transistor into the output. This will nearly always destroy the series device, since no simple fuse could operate rapidly enough in practice to rule out this type of failure. The simplest answer to this problem is to incorporate some rapid acting current-sensing circuit in the feed path to the pass device, which can be arranged either to 'steal' the input current to this transistor, as shown in FIG25a or to disconnect its drive, as shown in FIG25b. Frequently, such a current limiting circuit will be arranged to be 're entrant' in its operating characteristics, as shown graphically in FIG26, so that the output current which the pass device is allowed to deliver is related to the voltage existing across this device, which will limit the thermal dissipation of this device in the event of a sustained short circuit. This will also prevent the possibility of 'secondary breakdown' -- a failure mechanism which operates if the boundaries of the permitted area of current flow and collector voltage, shown for a bipolar power transistor in FIG27, are exceeded.

Practical series regulator circuitry

Some simple series voltage regulator circuits, which could, if necessary, incorporate current limiting systems, are shown in FIGs 28-31. All of these require some sort of voltage reference arrangement with which the output voltage from the regulator circuit can be compared. The simplest traditional type of layout used for this purpose is that shown in FIG28a, where an amplifier transistor, Qh is connected with its base fed from a potential divider (RVX and/^), connected across the output of the regulator, while the emitter of Qj is held at a fixed potential by the zener diode, D^ If the output voltage tends to increase, the current flow through R2 will increase, and the base voltage of Q2 will be lowered: an action which will reduce the regulator circuit output voltage again.

FIG25 Over-current limiting circuits Umax = 2A)

FIG26 Re-entrant over-current protection

FIG27 Typical power transistor safe operating area rating (SOAR) . Collector-emitter voltage (V)

FIG28 --- a. Simple series regulator circuit ; c. Conventional series stabilizer circuit



FIG31 Elaboration of circuit FIG30 to provide re-entrant current limiting.

The most immediate practical problems with this layout are that, while the output voltage can be adjusted by altering RVX, the lowest possible output voltage is V(£>{) + Vbe, (Q^, and it’s not possible, with any value of RVX, to adjust the output voltage down to zero. This circuit can be further simplified by dispensing entirely with the zener reference source, as shown in FIG28b, where the forward base-emitter potential of Qx is used as the reference. This arrangement works quite well, and allows the output voltage to be adjusted down to about 0.6V, but its regulation characteristics are poor because output voltage errors are also attenuated by the output potential divider (RVl and R!), which limits the ability of the circuit to correct these.

The output voltage control performance of the circuit can be improved by elaborating the input error comparator into a long-tailed pair configuration, as shown in FIG28c, where the zener diode has only to supply the base current of Q1, and the emitter impedance of Q1 is lowered by the emitter follower action of Q3.

All of the circuits shown in FIG28, suffer from the snag that the total base current required by Q2, which …can be quite large, must be supplied by R3. This leads to a voltage drop across R3, which, together with the necessary base-emitter potential for Q2, determines the minimum drop-out voltage for the circuit. If R3 has a large resistive value, this drop-out voltage will be large. If, on the other hand, R3 is made small to lessen this problem, the gain of the amplifying transistor, Q1 will be lowered, and the output voltage regulation factor, and the output impedance of the circuit, will be worsened. Moreover, at low or zero output currents from the circuit, all of the current flowing through R3 must be drawn off by Qj, so the current handling capacity of Q1? and its thermal dissipation characteristics, must be adequate.

In the case of the simpler circuit of FIG28a, under zero output current drain, the bias current not required by Q2 must also pass through the zener diode, so it too must have adequate current handling capability.

The use of a Darlington connected pair, as shown in FIG28c, will reduce the current handling capacity necessary for the control voltage handling stages, which will improve their gain, and the performance of the circuit. The snag, here, is that this will also tend to increase, somewhat, the drop-out voltage of the pass transistor.

An alternative layout, using the pass transistor in its inverted mode, is shown in FIG29. In this circuit, the input potential divider, RVXIRV is taken to a negative voltage reference source ( R2,01,C1), so that, if the output voltage from the regulator circuit is inadequate, Q1 is cut off, and the current flow through R3 causes both Q2 and Q3 to be turned hard on. Since the output voltage under these conditions will exceed that which is required, the potential now present at Q1 base will cause Q1 to conduct, and draw of f current which would otherwise flow through R3 into the base of Q3, causing the output voltage to be held at the required level.

Although the pass transistor is now operated in its common emitter, rather than its common collector (emitter follower) mode, and therefore offers a relatively high output impedance, all three of the transistors are operated as amplifiers, which gives a high regulator circuit open-loop gain, which reduces the circuit output impedance.

Advantages offered by this type of circuit are that the drop-out voltage can be very low, since the forward bias for the pass device is drawn from the OV line; that the zero output current flow through Q3 (and through the base-emitter path of Q2) will be zero, so that the dissipation of Q3 needs only to be that required by the base current of Q2 to provide the regulator circuit output current; and that the output voltage can be adjusted down to some 0.6V, as in FIG28b, but without the penalty of the relatively poor regulation factor offered by this layout.

An elaboration of this layout, due to the author (JLH. Wireless World, Jan. 1975, pp. 43-45), uses an operational amplifier in place of Qh which allows a further improvement in the precision of the output voltage control, and is shown in FIG30. In this the output voltage is adjustable down to within a few millivolts of zero volts, and the drop-out voltage -- the minimum practicable regulator voltage drop between supply and output potentials -- is less than IV. This circuit also uses Q1 in a current limiting arrangement, in which, on excessive output current demand, the power supply output voltage is progressively reduced, making it operate as a constant current source, but with the same degree of output noise and ripple rejection as it would possess in its normal constant voltage mode. The circuit can be made to operate in a re-entrant mode, if required, by connecting Q1 emitter to a tap on a further potential divider, as shown in FIG31, or, in the case of a more conventional circuit, as is shown in FIG32.

Voltage reference systems:

The use of a simple zener diode, or merely the forward base-emitter turn-on voltage for a junction transistor, have been employed, in the circuits so far described, as the reference voltage against which the regulator output voltage is compared to establish the required output level. All of these, however, are somewhat temperature sensitive, though it’s possible to temperature compensate an avalanche-type zener diode (i.e. one having a breakdown voltage in excess of some 6V), which will have a positive temperature coefficient, by connecting it in series with one or more forward biased silicon junction diodes, which will have a negative temperature coefficient. Indeed, temperature compensated zener diodes, such as the 1N821/1N827, are available, with a typical reference voltage of 6.2V, and a temperature coefficient as low as 0.001 %/°C. An alternative approach, and one which is used to provide a temperature stable voltage reference in al most all of the integrated circuit voltage stabilizer blocks, is to use a 'band-gap' arrangement.

FIG32 Elaboration of circuit of Figure 1528b to provide re-entrant current overload limiting.

The band-gap voltage reference circuit:

The use of a temperature compensated zener diode, with an output voltage of, say, 6.2V, would lead to some design difficulties in voltage stabilizers which were required to have an output voltage of, say, 5V. Also, all zener diodes have a significant noise component in their output voltage, and this would also appear in the output voltage of the regulator. For these reason s, the voltage against which the regulated output voltage is compared is normally derived from a band gap reference system. This type of circuit derives its name from the energy-band gap of the semiconductor material, at 0°Kelvin, which, for silicon, is 1.205V.

If, as in the circuit of Figure 15/33, two identical transistors, Q1 and Q2, are operated at substantially different currents -- perhaps so that that through Qj is ten times greater than that through Q2 -- there will be a voltage developed across R2, in the emitter circuit of Q2, which is equal to the difference between these two base-emitter voltages (defined as AVbe). This potential difference has a positive coefficient of voltage against temperature, and provides a means for generating a reference voltage which has a near zero temperature coefficient. The way by which this can be done can be seen from Figure 15/33. If the current gain of Q2 is sufficiently high that its base current flow can be neglected, the voltage developed across R3 will be ( R3/ R2)??^6. The total voltage across the circuit, from Q3 collector to the OV. line, will be, if Q3 is identical to Q1 and Q2, Voui = VbQ + (R3/R2Wbe and if these two component voltages -- one with a positive and one with a negative temperature coefficient -- add up to the band-gap potential of 1.205V, then the output voltage will have a zero temperature coefficient.

FIG33 Band-gap voltage reference circuit

IC voltage regulators:

The requirement for a zero temperature coefficient reference source is particularly critical in IC voltage regulators, since the voltage reference circuit will be contained within the IC package, and this package may become hot in use. A further point is that since all the silicon junctions are forward biased they will generate very little electrical noise. Also since the breakdown voltage of a zener diode is critically dependent on the doping level, and a selectively doped region of an IC chip would be inconvenient to arrange, and difficult to control to the required degree of precision, from the point of view of IC manufacture the more predictable behavior of circuitry using only forward biased junctions is much to be preferred.

The actual circuit layout used in most positive out put IC voltage stabilizers is of the general form shown in FIG34 -- which can be compared with the discrete component layouts illustrated in FIG28.

However, various circuit artifices are used to obtain performance refinements, such as the use of a low voltage drop constant-current source to feed the Darlington-connected output pass transistor pair to reduce the drop-out voltage, and to reduce the regulator IC output current if the chip temperature approaches its maximum permitted working level.

FIG34 Typical +ve input, IC voltage regulator schematic circuit

FIG35 Circuit layout ofLM109 positive voltage regulator

For those who are interested, a detailed explanation of the way in which these internal circuit arrangements operate in a typical IC voltage regulator is given in Wireless World, March 1982, pp. 41-44.

The complete circuit of the LM 109 positive output three- terminal voltage regulator IC is shown in FIG35. Although this IC is designed to provide a fixed +5V output, its output voltage can be varied, above this figure, up to some +22V, by the use of the circuit shown in FIG36 -- an arrangement which is commonly used in adjustable voltage regulators, such astheLM317.

The use of adjustable output voltage regulator ICs in normal power supply applications, as distinct from their use as zener diode substitutes, has been made largely unnecessary by the availability of both +ve and -ve output fixed voltage regulators covering the range 5_24V, of which the most common types are those derived from the Fairchild 78xx and 79xx series.

These have now become so widely adopted as power supply components that discrete transistor voltage regulators are now only justifiable where output volt age and current requirements are outside the range available from IC systems.

FIG36 Method of adjusting output voltage of fixed voltage regulator IC

FIG37 Fairchild 79xx voltage regulator circuit

Because of the difficulty of fabricating good quality PNP transistors in ICs, the negative output voltage regulators, such as the µF79?? devices, employ a pass transistor layout which is similar to that shown in FIG29, with the output taken from the pass transistor collector. The band-gap voltage reference circuit still uses NPN transistors, referred to the -ve line, with a DC amplifier (Q6, Qn , Q14 and Q17) employed to invert the DC control voltage for application to the pass devices. The circuit of the µF79?? voltage regulator is shown in FIG37. The use of the modified pass transistor configuration gives the 79xx type regulator an improved drop-out voltage (1.1 V) as compared with that of the 78xx positive rail regulator IC. The output noise figure (125-375µ\0 is, however, worse than that of the 78xx device (40 90µF), with the lower output voltage ICs having better noise figures than the higher output voltage equivalents. If a very low output noise figure is desirable, it’s worth remembering that the low output current ICs, such as the 78Lxx and 79Lxx devices have a noise output of only about one third that of their higher power equivalents.

Increased voltage and current ratings for IC regulators

The normal IC voltage regulator has an input voltage limit of 35V -- though the 24V output µF7824 and µF7924 ICs allow a 45 V maximum input voltage and there are some higher voltage devices available -- and this may be inadequate for some applications. In this case, the easiest way to improve the voltage and cur rent ratings is by hybridizing the IC with a few external discrete components.

For example, the maximum current rating can be extended by causing the IC to control a parallel connected power transistor, as shown in FIG38.

This layout takes advantage of the fact that an IC voltage regulator draws very little current when its output voltage slightly exceeds the predetermined out put voltage level. So by connecting the compound emitter follower circuit, Q1 ,Q2, in parallel with a resistor (R{) in series with the IC, the parallel high current transistor will only be forced into conduction when the IC input current reaches a high enough level to cause a 0.6V drop across Rv The output voltage can be adjusted, if necessary, by inserting a network R4, RV^, in the reference limb of the IC. This gives almost exactly the same precision of output voltage control, and input noise and ripple rejection as would be pro vided by the IC on its own. Over-current limiting can be provided for this circuit in exactly the same way as shown, in FIG25a, for a discrete component system.

A circuit allowing both a higher input and a higher output voltage to be controlled by a fixed voltage IC regulator, such as the µF7815, is shown in FIG39. In this, the input voltage to the 7815 is control led by the potential divider circuit ( R5, Re), and the emitter follower, Q2, while output current (and input current) is only drawn from the IC if the emitter voltage of Qj falls below 15V, which allows the output voltage to be adjusted down to about 14.5V by the setting chosen for RVV Over-current limiting is pro vided by the transistor Q3, which steals the drive current from the pass transistor if the voltage across Rd exceeds some 0.6V.

FIG38 Method of extending output current capacity of IC voltage regulator

FIG39 Circuit providing increased current and voltage capability from IC regulator

Switching and Switch-mode voltage regulators:

All of the voltage regulators examined so far have come within the category of DC to DC converters, in that they have taken a DC input; usually neither particularly clean, in terms of freedom from electrical noise and mains induced ripple, nor stable in voltage; and converted this into a stable, much more noise free, but rather lower output voltage, since the regulator required some internal voltage drop in order for it to operate. There are, however, other useful arrangements which allow, for example, an input source of one polarity, say +15V, to be converted into an output of the opposite polarity, say -15V, or, alternatively, which will allow a lower voltage, say +5V, to be converted into a higher voltage, say +15V. Since all of these systems rely on the transfer of the energy stored in a capacitor or an inductor from the input to the output circuit, by some repetitive switching action, these circuits are generally called 'switching regulators'.

There is also a specific category of switching power supply system in which the input DC source is just an input rectifier/reservoir capacitor circuit. This pro vides a high power, but rather rough DC supply straight from the raw AC line, and this is then converted into a stable DC output, isolated from the mains input, by a high frequency oscillatory switching circuit, coupled to a small high frequency transformer.

This type of circuit is employed because great savings in size, weight, and heat dissipation can be made by the use of high frequency, rather than 50-60Hz, volt age transformation techniques. The term 'switch mode' is usually used when referring to systems of this kind.

DC to DC reverse polarity converters

One of the simplest DC-DC converter systems, for which circuits frequently appear in the 'Design Ideas' columns of electronics magazines, uses a free-running square-wave generator, such as, say, a 555 type IC, connected as shown in FIG40, to produce an output in which the charge stored in C1 on each +ve going half cycle, can be rectified by the diodes D1 and D2, and 'pumped' into the output circuit. By the choice of the polarity of D1 and D2, the DC output from the circuit can be either +ve or -ve.

FIG41 Switching polarity inverter circuit

FIG42 Simple switch-mode regulator circuit

FIG40 Negative voltage generator circuit

A more elegant switched-capacitor converter circuit, using the same philosophy, is that shown in FIG41, based on a MC7660 IC. This gives a polarity-reversed output voltage which is very close to that of the input, for supply voltages up to 10.5V, and for output currents up to 10mA. The overall conversion efficiency of this circuit is typically about 95%. Similar types of device are available from Texas Instruments, National Semiconductor, and many other manufacturers.

A point which must be remembered in any switching regulator is that it’s capable of injecting a significant amount of noise back into its own supply line, so the voltage feed to the circuit must be adequately decoupled.

A much better type of circuit is the kind in which the energy store is an inductor, in that, for a given physical size, the current transferred to the output can be much greater. A typical example of this kind of arrangement is shown, schematically, in FIG42, in which a switching transistor Q1 is used, alternately, either to permit or to interrupt the current flow from the input DC supply through the inductor, Lv to the 0V line.

When the current flow through Lx is interrupted, the voltage at the collector of Q1 will rise, very rapidly -- a phenomenon known as the 'back EMF', or, more popularly, as the 'flyback effect'. This causes the diode, Dv to conduct, and causes current to flow into the output load, RL. The output voltage provided by the circuit is then compared, through the voltage divider network, RVVRX, with a reference voltage, and the amplified difference used to control the mark-to space ratio of a high frequency oscillator. This, in turn, controls the switching characteristics of Q1? to regulate the output voltage.

Although bipolar junction transistors can be used, as Qj, in this application, power MOSFETs are more commonly used because they have a more rapid switching action and their high input (gate) impedance makes them easier to drive. By reversing the positions of Lj and Q1? as shown in FIG43, so that when the current flow through Lx is interrupted the voltage at its live end suddenly swings negative, the output polarity can be reversed, to allow a -ve output voltage to be derived from a +ve input supply line. As in FIG42, the output voltage can be controlled, quite precisely, by comparing the output level with a reference source, and using the amplified difference to control the on-to-off time duration (the mark-to- space ratio), of the switching transistor, Q1.

FIG43 Negative output voltage switch-mode regulator circuit.

Since the switching frequency can be made quite high -- say 100kHz -- the value of the output smoothing capacitor, C2, does not need to be very high to reduce the residual switching ripple to the required low level.

Both of these switched inductor circuits can provide an output voltage which is greater than the input supply line level -- say, -15 V out from a +5 V in -- and typical efficiencies can be as high as 90-95%, so that the system runs cool.

The control of the mark-to-space ratio of the wave form used to operate the switching transistor, in order to adjust the output potential provided by the circuit, can be done by the kind of circuit shown in FIG44. In this a free-running 'triangular' or sawtooth waveform generator is 'sliced' by a high gain voltage comparator IC, which gives an output waveform in which the relative durations of the +ve and -ve parts of the cycle can be altered with reference to one another. If necessary, this output waveform can then be sharpened up still further, to improve the efficiency of switching, by reducing the length of time occupied in the on to off transitions, by interposing a CMOS logic gate IC between the output of the comparator IC and the switching device.

Sawtooth generator:

Mark-to-space * control voltage; Voltage comparator

FIG44 Adjustable mark-to-space ratio oscillator circuit

FIG45 Complete AC-DC switch-mode power supply

Switch-mode power supply (SMP) systems

A typical layout for a switch-mode power supply is shown in FIG45. In this, the mains line AC voltage supply is rectified by the diode bridge (D^, and charges a capacitor (C^), to provide a DC source of either 163V or 340V, depending on whether the input supply line is 115 or 240V. Either this, or some alternative DC supply is used to power a pulse-width modulated oscillator, operating normally in the range 25kHz-200kHz, depending on the designer's choice, which controls the conduction of the switching transistor Qj. In the past, this would normally have been a bipolar junction device, but the growing availability of higher switching efficiency MOSFETs has made these the more attractive choice.

It’s nearly always essential to provide full electrical isolation between the mains input and the DC output lines in any power supply, for safety in operation, and to allow freedom in the choice of interconnections in the load circuit. In the case of an SMP supply, this isolation is given by the transformer Tx. Because of the high operating frequency of this system, this trans former need only have a small core cross-sectional area -- in practice this core will often be a 'ferrite' toroid -- and the windings need only few turns, so that they can be wound with relatively thick wire which gives a low winding resistance and a high overall transformer efficiency.

The rectified output voltage is used to power a DC voltage reference source, against which the output DC level can be compared. As in the previous systems, shown in FIGs 42 and 43, the error voltage detected by the amplifier (Aj) is used to operate a pulse-width modulator circuit which controls the conduction cycle of the switching transistor, Q1# However, in the case of an SMP, the control circuit must be chosen so that a further isolation element, usually either an 'opto-coupler' or a small pulse transformer, can be interposed in the control path between the error amplifier and the switching transistor.

The DC operating supply for the switching oscillator, and other low voltage electronic circuitry shown on the left-hand side of the isolating transformer, is most easily provided by a simple, very low power, transformer-rectifier circuit of conventional form, al though various ingenious schemes have often been used in commercial designs to provide a short burst of DC to this circuit, on switch-on, to allow the oscillator power supply, once the oscillator has started working, to be derived from a tertiary winding on T^ Once again, since the 'ripple' frequency of the rectified supply is so high, only a small value reservoir capacitor (C 2) is needed to reduce this to an acceptable level. However, since the impedance of an inductor increases with frequency (ZL = 2 pi fL), even a low value inductor, with a very low DC resistance, interposed between the secondary rectifier bridge (D2) and the output capacitor (C3) and load, will give a substantial reduction in the output noise and ripple.

Because the SMP is so compact in size, and has such a low heat dissipation, this kind of power supply is now very widely used in 'desk-top' personal computers.

Also, because of the popularity of this type of circuit, both pulse-width modulator systems, and switched capacitor and inductor converters, as well as complete switch-mode regulator circuits are available in IC form, from a wide range of manufacturers, such as Harris, Hitachi, National Semiconductors, SGS, Siliconix, Silicon General and Texas Instruments.

Prev. ------- Next

top of page  Article Index  Home

Home | Glossary | Books | Links/Resources
EMC Testing | Environmental Testing | Vibration Testing

Updated: Wednesday, 2016-08-10 21:18 PST