HISTORY AND DEVELOPMENT OF SWITCHED-MODE POWER
SUPPLIES PRE 1987

Contents

1 Introduction

With the introduction of the transistor in the early 1950's, and especially with the development of integrated circuits from the early 1960's onwards (16), designers of electronic equipment, computers and instrumentation have increasingly demanded smaller, more efficient power sources to supply their equipment. Therefore, to meet this demand, the power supply itself has become more and more sophisticated. In fact the developments in power supply technology can be directly linked to the introduction of various power semiconductor devices, even though the theory, in many cases, was already well known.

The regulated power supply technology can really be divided into two distinct forms; firstly, the linear or series regulator and, secondly, the switched-mode conversion technique. Switched-mode technology is multi-facetted with a wide variety of topologies achieving the end result of providing a regulated DC voltage.

The main differences between the linear and switched-mode regulator are in the size, weight and efficiency. The linear regulator, Figure 1, utilises simple techniques of controlled energy dissipation to achieve a regulated output voltage independent of line and load variation. It is, therefore, inherently inefficient, especially when a wide input voltage range has to be catered for. When linear techniques are applied to regulating a low voltage from the mains 11OV or 24OV AC source, then the disadvantages of the technique become apparent.

Linear Regulator

Figure 1: Linear Series Regulator Equivalent Circuit

A typical linear power supply is shown in Figure 2. The step down, low frequency mains transformer is very bulky, large heat-sinking is required to dissipate the heat generated by the regulating element and very large filter capacitors are required to store enough energy at the voltage to maintain the output for a reasonable length of time when the mains source is removed. Switched-mode techniques, on the other hand, offer the possibility of theoretically loss-less power conversion. The switched-mode regulator, Figure 3, employs duty cycle control of a switching element to block the flow of energy and thus achieve regulation. It has the added advantage of when applied to off-line applications of giving significant size reduction In the voltage transformer and energy storage elements.

Practical Linear Regulator
Figure 2: Practical Linear Series Regulator Circuit

Switching Regulator
Figure 3: Switched-Mode Regulator Equivalent Circuit

Since a switched-mode converter can operate at significantly high frequencies, then a smaller transformer using ferrite cores can be used. Also since the high rectified mains voltage is chopped, then energy storage for hold-up can be accomplished on the primary side of the step-down transformer and so much smaller capacitors than the linear counterpart can be used. As shown in Figure 4 below.

Practical Switching Regulator
Figure 4: Practical Circuit of a Switched-Mode Regulator

Although the benefits of switched-mode techniques are great, there is a penalty paid in the increased noise present at both input and output of the supply due to the power switching techniques. Also the associated control circuitry is much more complicated than its linear counterpart.

Historically, the linear was very common during the late 1950's and early 1960's when power supplies using switching techniques were very rare. The ascendency of switched-mode power supplies is directly linked to the development of fast high voltage switching power transistors in 1967 (49) and, to a lesser extent, on developments in ceramic ferrite materials and capacitor technology. Even so, linear regulators and power supplies still have their areas of application to-day, such as low power below 50W, where the costs of the different technologies is comparable and also in stabilised bench power supplies. The linear is also preferred where a virtually noiseless supply is required.

2 Origins and the Development of Switched-Mode Techniques

The origins of switched-mode converters are linked with the developments in inverter circuitry. An inverter is a processor for generating AC from DC and is, therefore, a constituent of some forms of switched-mode power supplies. The earliest described inverters were developed before the first transistors appeared and, therefore, employed valves as switching elements, such as a push-pull inverter described by Wagner (1). After the first bipolar junction transistor was produced in 1948 (2), there was a proliferation in inverter/converter circuit designs. In 1952 a low power high voltage DC supply was described by Bryan (3) for applications such as Geiger counters. This utilised a transistor in an oscillator with power delivered to the output in a flyback mode via a transformer.

From 1952 onwards, Germanium power transistors increasingly became available (26,) so giving impetus to inverter/converter developments. The various forms of transistor switching circuits developed during the 1950's were categorised into three main groups by the end of the decade (11), namely Ringing Choke, Self-Oscillating, Push-Pull and Drive Push-Pull Converters.

The Ringing Choke Single-Ended Converter, illustrated in Figure 5, employs a single transistor in a transformer coupled relaxation oscillator which turns the transistor on and off periodically. Energy is transferred to the output during the switch off stroke and so its modern equivalent is the Flyback Converter. The circuit was only suitable for low power applications and was mainly used to supply a DC voltage. A paper by Light and Hooker (6) is a well-known reference on Ringing Choke Converters, dealing with all aspects of the circuit including methods of feedback control and regulation.

Ringing-Choke
Figure 5: Ringing-Choke Single-Ended Converter

The Self-Oscillating Push-Pull Converters are found in the majority of medium power applications, 10-100W, although a 250W converter is described in a paper by Uchrin (8). The circuit uses a pair of transistors in a symmetrical square-wave oscillator. The transistors switch on and off alternately, connecting the DC input voltage to one or other side of a centre-tapped transformer winding. This inverter produces a square-wave voltage across the transformer secondary which can be rectified to produce a DC output voltage. There are two main methods of causing the circuit to oscillate. The basic form shown in Figure 6 employs a feedback winding on the main transformer, and has been described by Royer (7) and Uchrin and Taylor (4). The two transistors' conduction alternates when the transformer, usually made of square hysteresis loop core material, saturates. Other self-oscillating forms, Figure 7, employs a separate feedback drive transformer which controls transistor switching. Such a circuit is described in a paper by Jensen (9).

Single Transformer Oscillator
Figure 6: Single Transformer Self-Oscillating Push-Pull Converter

Two Transformer Oscillator
Figure 7: Two Transformer Self-Oscillating Push-Pull Inverter

Driven Push-Pull circuits, Figure 8, are more commonly used in high power applications, where the frequency stability over self-oscillating forms is an advantage. The push-pull circuit is driven by a separate master oscillator which controls the frequency of operation and can be used to supply either DC or AC depending on the presence of output rectifier or otherwise. Of all the early forms of power switching circuit, it is the driven push-pull which is still commonly used as part of some power supplies at the present day.

Push-Pull
Figure 8: Driven Push-Pull Converter

The 1960's heralded the development of the modern forms of switching regulators and switched-mode power supplies. During the early 1960's three forms of non-dissipative switching regulators were developed for low voltage DC to DC applications. They are the buck, boost and buck-boost (14) regulators. The buck regulator steps down the input voltage to a lower regulated output voltage.The boost regulator steps-up the input voltage to a higher regulated level. The buck-boost regulator, also referred to as a flyback regulator, is used to regulate a negative voltage at a level higher or lower than the positive input voltage. The method of regulator control in all cases is achieved by varying the duty ratio of the electronic switch, most commonly by pulse width modulation (PWM). These switching regulators will be dealt with further in Section 3.

The main drive in switched-mode converter developments was the aerospace industry (15, 16), where the advantages of small size and high efficiency are of great importance. Power supplies working from the mains for the commercial market were still largely linear forms, with isolation, a necessary requirement for safety, being achieved by using a bulky mains frequency transformer. The major breakthrough for the commercial power supply section occurred at the end of the 1960s with the introduction of high voltage silicon switching transistors (19, 26).

The original impetus for the development of these devices came from the television industry. Power supply design engineers were now able to implement switching regulator and inverter technology into stabilised power supplies operating from the mains. In the USA, with the advantage of a lower 110V rms. mains supply than Europe, manufacturers of power supplies were the first to introduce switched-mode converters, notably Trio. In Europe too, switched-mode power supply developments took place using the new high voltage switching transistors.

One of the first commercially available ranges of switched-mode power supplies were the 'MG' single output units introduced by Advance Electronics Ltd. in 1972 (25). Several other companies also entered the market soon afterwards with their own switched-mode power supply ranges, including Farnell Instruments Ltd. The converter topology favoured in these first designs was the half-bridge circuit controlled using pulse width modulation, Figure 9. This topology enables the best utilisation of switching device in terms of voltage rating for a two transistor system. The voltage rating of switching transistors need only be greater than the peak of the input AC voltage, since high voltage transistors above 500V rating were not readily available (31).

PWM Half-Bridge
Figure 9: Functional Diagram of a PWM Half-Bridge Converter

3 The Modern Switched-Mode Power Supply Topologies and Trends

3.1 The Switching Regulator Family

While the commercial switched-mode power supply manufacturing industry was beginning to grow during the 1970s, the theory and technology of switched-mode conversion was being relationalised as part of the academic discipline of Power Electronics. By far the greatest contribution made within the discipline has been the work done by R.D. Middlebrook and his colleagues in the Power Electronics Group of Caltech in California, USA. The initial work of the Caltech Group, started in 1970, was aimed at developing models for the three basic dc-to-dc switching regulator topologies already developed in the 1960s, namely the buck, boost and buck-boost converters (27, 29). From this work stemmed the modelling and analysis method called state-space averaging (31, 34) which allowed the theoretical prediction of a converters frequency response, and therefore a better understanding of a switched-mode regulator's feedback loop and stability criteria. Further work at Caltech, especially by Cuk in his PhD Thesis, produced a fourth member of the basic dc-to-dc switching regulators which has been described as an optimum topology (35) because of its symmetrical structure and non-pulsating input and output currents. The new optimum topology dc-to-dc switching regulator is now commonly known as the Cuk converter after its inventor and completes the family of single-switch non-isolated switching regulators. The family are shown in Figure 10 together with the steady-state current waveforms and transfer equations.

Buck Regulator
(a): Buck Regulator

Boost Regulator
(b): Boost Regulator

Buck-Boost Regulator
(c): Buck-Boost Regulator

Cuk Regulator
(d): Cuk Regulator

Figure 10: The Switching Regulator Family

3.2 Isolated Topologies

In many applications, such as operating off the AC mains, isolation is a necessary requirement within the converter between input and output. By inserting an isolation transformer into the four basic switching regulator topologies, then the four single-ended isolated switching converters are derived as shown in Figure 11

Forward Converter
(a): Forward Converter

Isolated Boost Converter
(b): Isolated Boost Converter

Flyback Converter
(c): Flyback Converter

Isolated Cuk Converter
(d): Isolated Cuk Converter

Figure 11: The Single Transistor Isolated Topology Family

The isolated buck and isolated buck-boost topologies are more commonly referred to as the forward and flyback converters respectively, and are the most popularly topologies used in switched-mode power supplies manufactured commercially.

3.3 Multiple Switch Topologies

The main disadvantage of the single switch topologies is the need for the high voltage blocking capability of the transistor switch (twice the DC input voltage), especially when operating from a rectified AC mains supply. Also the single switch topology is not an ideal solution for higher power converters, where the current rating of the transistor switch needs to be much greater. Therefore another group of isolated converters utilising more than one switch can be identified. Figure 12 illustrates the three multiple switch topologies, namely the half-bridge, full-bridge and push-pull converters. All three converters are buck derived due to the nature of switching involving pulsating input current and non-pulsating current and also having an identical ideal voltage gain of the forward converter.

Push-Pull Converter
(a): Push-Pull Converter

Half-Bridge Converter
(b): Half-Bridge Converter

Full-Bridge Converter
(c): Full-Bridge Converter

Figure 12: The Multiple Switch Topology Family

The original self-oscillating push-pull converters described in the 1950s by Royer and Jensen are really members of this group. These topologies also have an added advantage over the single-ended forward and flyback converters in that full flux excitation of the transformer core occurs instead of only a half core flux capability. This makes these multiple switch topologies more suited to higher power operation.

3.4 Alternative Topologies

The single-ended and push-pull groups of SMPS represent the basic, most common converter topologies used in commercial applications from the mid-seventies up until the present day. These "square-wave" switching converters do have drawbacks when their design is realised in practice. Switching Bipolar Junction Transistors (BJTS) do not turn-on and off immediately but do exhibit finite rise and fall times in their switching behaviour (48) which can lead to a number of undesirable affects in conventional converter topologies.

In multiple switch converters, transistor switching overlap can occur which could cause catastrophic failure to the converter by effectively applying a short circuit to the supply source. A number of researchers have worked on alternative topologies and protection circuitry to eliminate this problem. One such alternative topology worthy of note is the Weinberg push-pull converter (28), Figure 13, which is inherently self-protecting when there is any possibility of component imbalance or conduction overlap.

Weinberg Converter
Figure 13: Weinberg Push-Pull Converter

In all basic switched-mode topologies, the finite duration of the switching transitions will cause a high peak pulse power dissipation in the device, as shown in Figure 14. This produces a degradation in converter efficiency and worst of all, can lead to transistor destruction during the turn-off transition due to the inherent BJT second-breakdown phenomenon. Therefore the greatest amount of research into alternative switched-mode topologies has been in the field of resonant converters. These converters have tuned circuits as part of the power conversion stage and exhibit sinusoidal voltages and/or currents, so leading to transistor switching transitions at the ideal conditions of zero stress. The development of Resonant conversion is discussed more fully in a separate document.

Switching Losses
Figure 14: Switching Losses in Power Transistors

3.5 Control Methods

In the majority of converter topology applications, Pulse Width Modulation is used for controlling the converter's output voltage through feedback control of the switching transistor.Other forms of control are becoming increasingly popular. One such technique is current-mode control (45) which utilises the switching transistor current as a control parameter and has the benefit of providing an inherently more stable closed loop response. Another control method finding favour with power supply designers is feed forward (30) which improves the transient load and line response of mains driven power supplies.

3.6 Switched-Mode Topology Applications

The switched-mode power supply market is now well established within the electronics sector (43, 44), with a large number of power supply manufacturers world-wide providing a wide range of units for the commercial and military markets. The main end-user systems for switched-mode supplies, are computers, both large mainframe and smaller, personal and word processors, and the various telecommunications systems. A typical system often requires a number of output voltages from its power supply and therefore the majority of power supplies tend to be multiple output forms. Typical supply voltages are +5V for Bipolar logic, +2V, -5V, for ECL logic, +12V for CMOS logic, +12V, +15V for operational amplifiers, and +24V for DC motors such as disc drives.

The topologies and control methods used to achieve the desired output voltages in the various power ranges tends to vary from manufacturer to manufacturer. The topologies reviewed previously in this section all have their more favoured applications. In general switching regulators are usually used as secondary regulators on multiple-output units, Isolated single-ended configurations are used in low power single or multiple output ac to dc converters and multiple switch topologies are used for higher output power applications. Also used as secondary regulators in some multiple output power supplies are linear regulators, mainly three terminal integrated circuits in low current outputs and magnetic amplifiers (21) for higher current outputs.

Overall, the various converter topologies can be arranged in an applications 'tree' as illustrated in Figure 15.

Family Tree
Figure 15: Power Converter Family Tree

3.7 Future Trends

Most commercial switch-mode power supplies in the market to-day operate in the frequency range 10kHz to 50kHz (43). There is now a growing trend in research work and new power supply designs in increasing the switching frequencies upwards to 100kHz and above (36). The reason being to reduce even further the overall size of the power supply in line with miniaturisation trends in electronic and computer systems. Several new developments in circuitry techniques and components have helped the drive towards higher operating frequencies and size reduction.

The introduction of the power Metal Oxide Semiconductor Field Effect Transistor (MOSFET) (47, 49), at the end of the 1970s, has provided the power electronics industry with a device capable of switching at much greater speeds than BJTS. This is due to the MOSFETs' inherent lack of storage and fall time affects when turned off. Therefore MOSFETS are now increasingly replacing BTJs in new designs operating at much higher frequencies. However, MOSFETs still have their limitations especially in voltage rating and cost. The device has a much higher fabrication cost compared with BJTs due to it being more akin to the integrated circuit than a discrete component. This has prohibited its application where the component cost of the power supply is paramount. The intrinsic characteristics of the MOSFET produce a large on-resistance which increases excessively when the devices' breakdown voltage is raised. Therefore, the power MOSFET is only useful up to voltage ratings of 500V and so is restricted to low voltage applications or in two-transistor forward converters and bridge circuits operating off-line. Improvements in fabrication techniques and device characteristics are still in progress and so the MOSFET is likely to replace BJTs in most applications especially as the cost per device is reduced.

Another new device likely to displace the BJT in many high power applications is the Insulated Gate Transistor (IGT) (49). This device combines the low power drive characteristics of the MOSFET with the low conduction losses and high blocking voltage characteristics of the BJT. Therefore the device is highly suited to high power, high voltage applications. However, since current transport in the device is by the same process as the BJT, its switching speed is much slower than the MOSFET, so the IGT is at present limited to lower (<50kHz) applications.

The growth of the switched-mode power supply market during the late 1970s has stimulated many semiconductor component manufacturers to introduce special control ICs for switched-mode power supplies (38). The majority of these highly integrated control and regulating circuits contain the necessary constituents of the Pulse Width Modulator for a switched-mode power supply. Other control ICs offer the current mode control method or feed forward capabilities. These control ICs also include many protective and monitoring circuits such as current limit, overvoltage protection and soft-start-up. The implementation of a control IC within a power supply should reduce the development costs associated in designing discrete control circuitry and also save circuit board area occupied by the control and monitoring circuitry.

An extension of the integrated control circuit philosophy is the amalgamation of the same drive and control circuitry with a power device especially the MOSFET. In the future (49), more and more integrated power devices will be introduced so simplifying board layout and reducing component count.

Overall, the operating frequency of the conventional switched-mode topologies such as the half-bridge will be increased by improvements in components. Also developments in new topologies, especially resonant converters, will aid the move to higher operating frequencies and so reduce the overall size of the switched-mode power supply. Above all the speed with which these developments will take place will be governed by market forces within the commercial sector. The driving force in every manufacturer's design will always be the combined component and production costs. Therefore, any new device or topology will have to justify its implementation based on mainly commercial criteria.

Copyright (c) 1987, 1998 Steve Watkins