Voltage References and Voltage (part 2)

4.2 Converter Topologies

The converter topology is the arrangement of components within the converter. Converter topologies fall into two main categories: transformer-isolated and non-transformer-isolated. Each category contains several topologies, with some available in both forms.

4.2.1 Non-Transformer-Isolated Converter Topologies

This type of switching converter is used when some external component, such as a 50-60 Hz transformer or bulk power supply, provides the DC isolation and protection. These are simple, hence easy to understand and design. However, they are more prone to failure due to the lack of DC isolation. Therefore, these are used by designers mostly in situations such as distributed power systems, where a bulk power supply provides the necessary DC isolation.

There are four basic non-transformer-isolated topologies, the buck (or step down), the boost (or step up), the buck-boost (or inverting), and the Ctik converters. Each of these types is described in detail by Linear Technology Corporation, Datel Inc., Siliconix Inc., Motorola Inc., and Unitrode Integrated Circuits.

4.2.2 Transformer-Isolated Converter Topologies

In non-transformer-isolated converter topologies, only semiconductors provide DC isolation from input to output. Transformer-isolated switching converters rely on a physical dielectric barrier to provide galvanic isolation.

Not only can this withstand very high voltages before failure, it also provides a second form of protection in the event of a semiconductor failure. Another advantage is the ease of adding multiple outputs to the power supply without separate regulators for each. These features make transformer-isolated topologies preferred by designers.

The transformer-isolated category includes both forward and flyback mode converter topologies. The isolation transformer now provides a step-up or stepdown function. The topologies are called isolated forward, flyback, push-pull, half-bridge, and full-bridge. Each of these types is described in detail by Linear Technology Corp., Datel Inc., Siliconix Inc., Motorola Inc., and Unitrode Integrated Circuits.

4.2.3 Selection of a Converter Topology

The converter topology has a major bearing on the conditions in which the power supply can operate safely and the amount of power it can deliver. Cost versus performance trade-offs also are needed in selecting a suitable converter topology for an application.

The primary factors that determine the choice of topology are whether DC isolation is needed, the peak currents and voltages to which the power switches are subjected, the voltages applied to transformer primaries, the cost, and the reliability.

FIG. 13 illustrates the approximate range of usage for these topologies. The boundaries to these areas are determined primarily by the amount of stress the power switches must endure and still provide reliable performance.

The boundaries delineated in FIG. 13 represent approximately 20A of peak current.

FIG. 13 Industry favorite configurations and their areas of usage.

The flyback configuration is used predominantly for low- to medium-output power (150 W) applications because of its simplicity and low cost. Unfortunately, the flyback topology exhibits much higher peak currents than the forward-mode supplies. Therefore, at higher output power, it quickly becomes an unsuitable choice. For medium-power applications (100-400 W) the half-bridge topology becomes the predominant choice. The half-bridge is more complex than the flyback and therefore costs more. However, its peak currents are about one-third to one-half those exhibited by the flyback. Above 400 W, the dominant topology is the full-bridge, which offers the most effective utilization of the full capacity of the input power source. It also is the most expensive to build, but for those power levels, the additional cost becomes a trivial matter. Above 150 W, the push-pull topology sometimes is used. However, this exhibits some shortcomings, such as core imbalance, that may make it tricky to use. Estimating the major power supply parameters at the beginning of the design and using charts such as FIG. 13 will lead to the selection of a final topology that is safe, reliable, and cost effective.

Some relative merits of the converter topologies as well as their advantages, disadvantages, and typical applications are summarized in TBL. 5. Mathematical expressions for estimating some of the important parameters, such as peak currents and voltages and output power, are found in Brown (1990, 1994). A comparison of the strengths and weaknesses of the different converter topologies also is found in Moore (1994). The industry has settled into several primary topologies for a majority of the applications. Due to reasons discussed before, transformer-isolated topologies have become more popular than non-transformer-isolated topologies.

4.3 Control Techniques

Two basic methods of control are used in switching regulators: pulse width modulation (PWM) and resonant. In both techniques, the output voltage is sampled and the switching of the transistors is modified in some manner to keep the output constant.

TBL. 5 Comparison of Converter Topologies

FIG. 14 Voltage-mode control: (a) Block diagram, (b) Associated waveforms

4.3.1 Pulse Width Modulation Control Techniques

In a PWM switch mode power supply, a square wave pulse normally is generated by the control circuit to drive the switching transistor on and off.

By varying the width of the pulse, the conduction time of the transistor correspondingly is varied, regulating the output voltage. The major function of the control subsection of a PWM supply therefore is to sense any change in the DC output voltage and adjust the duty cycle of the power switches to correct for such changes.

An oscillator sets the basic frequency of operation of the power supply. A stable, temperature-compensated reference is used, to which the output voltage is compared in a high-gain voltage error amplifier. An error voltage-to-pulse width converter is used to adjust the duty cycle.

The PWM control circuit may be single ended for driving single-transistor converters, such as the buck or boost topology, or it may be double ended to drive multiple-transistor converters, such as the push-pull or half-bridge topology.

Two basic modes of control are used in PWM converters: voltage mode and current mode.

Voltage-Mode Control

This is the traditional mode of control in PWM switching converters. It also is called single-loop control, as only the output is sensed and used in the control circuit. A simplified diagram of a voltage-mode control circuit is shown in FIG. 14.

The main components of this circuit are an oscillator, an error amplifier, and a comparator. The output voltage is sensed and compared to a reference. The error voltage is amplified in a high-gain amplifier. This is followed by a comparator, which compares the amplified error signal with a sawtooth waveform generated across a timing capacitor.

The comparator output is a pulse width modulated signal that corrects any drift in the output voltage. As the error signal increases in the positive direction, the duty cycle is decreased; and as the error signal increases in the negative direction, the duty cycle is increased.

The voltage mode control technique works well when the loads are constant.

If the load or the input changes quickly, the delayed response of the output poses a drawback to the control circuit, as it senses only the output voltage. Also, the control circuit cannot protect against instantaneous overcurrent conditions on the power switch. These drawbacks are overcome in current mode control.

FIG. 15 shows a functional block diagram of the commonly used TL494 voltage mode controller. Control for both push-pull and single-ended operation can be achieved with this chip. The frequency of the oscillator is set by the external resistors Rr and Cr. The typical operating frequency is 40 kHz, and the maximum is 200 kHz.

Output pulse width modulation is accomplished by comparison of the positive sawtooth waveform across the capacitor to either of two control signals.

The NOR gates, which drive the output transistors, are enabled only when the flip-flop clock input is in its low state. This happens only during that portion of time when the sawtooth voltage is greater than the control signals. Therefore, an increase in control signal amplitude causes a corresponding linear decrease of output pulse width.

FIG. 15 A functional block diagram of the TL494 voltage-mode controller IC. (Motorola Inc.)

FIG. 16 Current-mode control: (a) Block diagram, (b) Associated waveforms

FIG. 17 A functional block diagram of the MAX747 current-mode controller IC. (Maxim Integrated Circuits.)

Current-Mode Control

This is a multiloop control technique, which has an AC current feedback loop in addition to the voltage feedback loop. The second loop directly controls the peak inductor current with the error signal rather than controlling the duty cycle of the switching waveform. FIG. 16 shows a block diagram of a basic current mode control circuit.

The error amplifier compares the output to a fixed reference. The resulting error signal then is compared with a feedback signal representing the switch current in the current sense comparator. This comparator output resets a flip-flop that is set by the oscillator. Therefore, switch conduction is initiated by the oscillator and terminated when the peak inductor current reaches the threshold level established by the error amplifier output. Thus, the error signal controls the peak inductor current on a cycle-by-cycle basis. The level of the error voltage dictates the maximum level of peak switch current. If the load increases, the voltage error amplifier allows higher peak currents. The inductor current is sensed through a ground-referenced sense resistor in series with the switch.

The disadvantages of this mode of control are loop instability above 50% duty cycle, less than ideal loop response due to peak instead of average current sensing, and a tendency toward subharmonic oscillation and noise sensitivity, particularly at very small tipple current.

However, with careful design, these disadvantages can be overcome. Therefore, current-mode control becomes an attractive option for high-frequency switching power supplies.

FIG. 17 shows a functional block diagram of the MAX747 current mode controller from Maxim Integrated Circuits. This is a CMOS step-down controller, which drives external P-channel FETs. The IC operates in a continuous mode under heavy loads but in a discontinuous mode at light loads. Stability of the inner current-feedback loop is provided by a slope-compensation scheme that adds a ramp signal to the current-sense amplifier output. The switching frequency nominally is 100 kHz and the duty cycle varies from 5 to 96%, depending on the input/output voltage ratio. EXT provides the gate drive for the external FET.

4.3.2 Resonant Control Techniques

The term resonance refers to a continuous sinusoidal signal. Resonant converters process power in a sinusoidal form. Resonant techniques have long been used with thyristor converters, in high-power SCR motor drives, and in UPSs. However, due to its circuit complexity, it had not found application in low-power DC-to-DC converters until recently. With the development of surface mount technology, resonant forms of DC-to-DC converters have gained rapid acceptance recently, harnessing certain advantages inherent in resonant techniques. The thrust toward resonant mode power supplies has been fueled by the industry's demand for miniaturization, together with increasing power densities and overall efficiency, and low EMI. With available bipolar devices and circuit technologies, PWM converters have been designed to operate with switching frequencies in the range 50-200 kHz. The advent of power MOSFETs enabled the switching frequencies to be increased to several MHz. However, increasing the switching frequency, although allowing for miniaturization, leads to increasing switching stresses and losses, which leads to a reduction in efficiency. The detrimental effects of the parasitic elements also become more pronounced as the switching frequency is increased. With quasi-resonant techniques, higher frequency as well as higher efficiency compared to PWM techniques are achieved.

Resonant circuits in power supplies operate in two modes that define the flow of current in the resonant circuit: continuous and discontinuous. In the continuous mode, the circuit operates either above or below resonance. The controller shifts the frequency either toward or away from resonance, using the slope of the resonant circuit's impedance curve to vary the output voltage. This is a truly resonant technique but is not commonly used in power supplies due to its high peak currents and voltages.

In the discontinuous mode, the control circuit generates pulses having a fixed on-time but at a varying frequency, determined by the load requirements.

This mode of operation does not generate continuous current flow in the tuned circuit, is the common mode of operation in a majority of resonant converters, and is called the quasi-resonant mode of operation.

FIG. 18 The resonant principle: (a) The basic circuit, (b) Associated waveforms

The Quasi-Resonant Principle

The quasi-resonant principle is used in power converters by incorporating a resonant LC circuit with the power switch. The power switch is turned on and off in the same manner as in PWM converters, but the tank circuit forces the current through the switch into a sinusoidal form. The actual conduction period of the switch is governed by the resonant frequency f r of the tank circuit. This basic principle is illustrated in FIG. 18.

The tank circuit exhibits a relatively fixed ringing period to which the conduction period of the power switch is slaved.

The on-period of the power switch is fixed to the resonance period of the tank elements. The quasi-resonant supply is controlled by changing the number of on-times of the power switch per second.

All resonant control circuits keep the pulse width constant and vary the frequency, whereas all PWM control circuits keep the frequency constant and vary the pulse width. The main advantages of the quasi-resonance techniques arise from the sinusoidal wave shapes of the switching currents and voltages.

The switching losses are reduced, leading to higher efficiency and greatly reduced EMI. The resonant switch consists of a semiconductor switch and resonant LC elements. Because resonant circuits generate sinusoidal waves, designers can operate the power switches either at zero current or at zero voltage points in the resonant waveform. Based on this, there are two types of resonant switches: zero current switches (ZCS) and zero voltage switches (ZVS), The two types of switches are the duals of each other. A description of the operations of these switches is found in Brown (1990). Comparable converter topologies as with the PWM technique are available with ZCS and ZVS quasi-resonant switching techniques. Circuits for each of these topologies are found in Brown (1990, 1994), Ctik and Middlebrook (1994), Steigerwald (1984), and Ctik and Maksimovic (1988).

Control Techniques

Control methods for resonant converters are variable frequency ones. Either the on-time is fixed and the off-time is variable or vice versa.

The basic functioning of a resonant mode control circuit is shown in FIG. 19. The fundamental blocks are a wide-band error amplifier, a voltage-controlled oscillator (VCO), and a temperature-stable one-shot timer.

The output voltage is compared with the reference, and the error voltage is used to drive the VCO. The VCO output triggers the one-shot timer, whose pulse duration is fixed as required by the convener. These control techniques are based on the voltage mode of control and, hence, suffer from poor input transient response characteristics.

Some representative control ICs currently available are the MC34066, LD405, UC1860, and UC3860. One of the first resonant mode controller ICs to appear on the market was the LD405, Subsequently, the CS3805 by Cherry Semiconductors and an improved GP605 by Gennum Corp. were released.

The LD405 and the CS3805 have drive current capabilities of 200 mA and operating frequencies in the range 10 kHz-1 MHz. They have single-ended and complementary outputs for driving power MOSFETs. The dissipation rate is specified at 500 mW at 50°C. The operating range of the GP605 is between 1 kHz and 1.2 MHz.

Unitrode Integrated Circuits introduced its UC3860 family of resonant mode controller chips in 1988. These could supply 800 mA, about four times the drive current, and operate at higher frequencies (up to 2 MHz) than the previously available ICs. Their power dissipation rate is specified at 1.25 W at 50°C. A newer series of resonant mode control chips by Unitrode Integrated Circuits and their applications are described in Wofford (1990). These are the UC3860, UC1861 (dual ZVS), UC1864 (single ZVS), and UC1865 (dual ZCS). A functional block diagram of the UC 1860 from Unitrode Integrated Circuits (1995-96) is shown in FIG. 20. The nominal operating frequency for this IC is 1.5 MHz, and the device implements resonant mode, fixed on-time control as well as a number of other power supply control schemes with its various dedicated and programmable features. The IC contains dual high-current-totem-pole output drivers that can be programmed to operate alternately or in unison (Unitrode Integrated Circuits, 1995-96).

FIG. 19 Resonant mode control

FIG. 20 A functional block diagram of the UC1860 resonant mode controller IC. (Unitrode Integrated Circuits.)

Although varying in detail and complexity, all these resonant mode control chips provide the same basic functions. All contain a VCO that varies the operating frequency, a monostable circuit that establishes the pulse on-time, and a steering circuit that determines the output drive mode (controlled on- or off- times, single-ended, or complementary). The chips also provide many of the same basic protection features such as soft start, undervoltage lockout, and overload protection although in different ways and with varying degrees of sophistication. Some of these control ICs are also described by Linear Technology Corporation (1990, 1992), Datel Inc. (1991), Siliconix Inc. (1994a), Motorola Inc. (1987), and Unitrode Integrated Circuits (1995-96).

4.4 Special Application Requirements

4.4.1 Sub-5-Volt Applications

As IC process lithography becomes finer and the need to reduce overall power dissipation becomes more important, ICs demand lower and better-regulated power supply rails. Power supply rails for modern microprocessors and logic ICs have dropped below 5 V to a 3.3 V pseudo-standard and are fast heading toward sub-2 V and sub-1 V levels.

These low-voltage supply requirements also differ in voltage level from system to system and from version to version of the same IC. If Intel's Pentium processor is taken as an example, some versions just require 3.3 V while others require two supplies at 3.3 V and 3 V. Yet other versions require supplies such as 3.383 V and 3.525 V, being the optimum voltages for maximum speed in the respective versions. Furthermore, the Pentium Pro comes with a 4-bit operating voltage code, which selects 1 of 16 discrete voltages between 2 and 3.5 V.

Lower voltage levels, however, don’t translate to low power. These multimillion-transistor ICs use CMOS technology, which at today's high operating frequencies consume substantial amounts of current although virtually no DC current. The power levels required therefore are rising. Top-of-the-line CPU power is approaching 100 W; and desktop CPU, power 30 W. Systems such as high-performance multiprocessor servers are approaching 400-600 W. The combination of increasing power requirements and reducing voltage rails requires the power source to provide and distribute very high currents within a system. For example, a 300 W system running off a 3 V supply must provide and distribute 100 A of current.

Furthermore, clock rates heading toward 400-600 MHz imply high current transients as devices come out of the sleep mode within a clock cycle of a few nanoseconds.

In addition to microprocessors, high-speed data buses, such as the 60 MHz GTL (Gunning Transceiver Logic) bus, are being recommended for interconnection of processors and peripherals on the motherboard. These require an active 1.2 or 1.5 V terminator at each end, each potentially capable of handling up to 7 A. Other data buses with similar requirements include Futurebus (2.1 V) and Rambus (2.7 V). Therefore, the challenges facing the power supply designer in the sub-5V range can be summarized as follows:

• Multiple voltage rails.

• Tight tolerances.

• High efficiency while generating high current (low power loss).

• High accuracy.

• High current transients.

The basic design requirements in the face of these challenges are low power loss and higher gain and bandwidth in the control loop to handle tighter voltage regulation and transients.

4.4.2 Converters for Battery-Operated Equipment and Battery Chargers

The proliferation of portable computers and handheld communication devices has opened up another branch in DC/DC converter evolution. Users of these portable devices demand features such as compactness, lightness, and low power consumption for extended periods of operation. Corresponding trends in DC/DC converters emphasize excellent conversion efficiency and compactness.

Additional constraints placed on the DC/DC converter include minimum noise intrusion into communication and audio circuits in the device. The increase in the integration of high-speed modems and CD-ROM drives into portable computers has made power supply switching noise an important consideration.

A key requirement for designers of battery-powered products is that they minimize the number of cells used in the product. Products ideally should run off a few high-capacity cells to minimize size and cost. Tiny devices such as pagers run off one or, at most, two cells. So do telephones and PDAs. Converters operating off input sources as low as that of a single-cell alkaline battery (1 V) are needed. Therefore, the voltage required, in most instances, is greater than the voltage available from the cells. Furthermore, for extended battery life, the converter should be able to operate from waning batteries.

Some systems need to operate with input voltages that approach the output voltage. This low-dropout condition requires the DC/DC converter duty cycle to approach 100%. Furthermore, a waning battery may swing the input voltage from a value above the output to one below it.

High efficiency is another prime consideration in battery-operated equipment. This means not only increased operating time on a battery charge but also reduced heat that must be dissipated in or removed from the IC and the device.

Low quiescent current is another requirement for extended battery charge.

Therefore, the challenges facing the power supply designer for battery-powered equipment can be summarized as follows:

• Low-dropout voltage.

• Extended use of battery charge.

• Small size.

• Low EMI.

4.5 Practical Design Approaches

This section describes-some of the approaches taken in DC/DC converter and controller design to address the special requirements of sub-5V and battery-powered applications. Illustrative examples of some successful solutions are presented.

4.5.1 Synchronous Rectification

In a conventional DC/DC converter such as the buck converter, typically an N-channel MOSFET with a low Ros(on) is selected for the switch. However, the diode's forward voltage becomes a limiting factor in improving the converter's efficiency, as the output voltage drops. This has led to the design of synchronous converters; these replace the diode, which normally is a Schottky, with another N-channel MOSFET. This usually is called the lower MOSFET, and the switch is called the upper MOSFET. The lower MOSFET conducts current during the off-time of the upper MOSFET. FIG. 21 shows a simplified diagram of a synchronous buck converter. In conventional synchronous converters, a single IC is used for PWM control and the synchronous drive of two external MOSFETs.

Newer process technologies take another approach, where a SynchroFET integrates the two MOSFETs, their drive circuits, and the synchronous control logic. This IC can be used with a conventional PWM control IC to design a converter with features superior to those with discrete MOSFETs (Maxim Integrated Circuits). Also, this approach allows the SynchroFET to be paired with many different PWM controllers to achieve various performance trade-offs.

HIP5015 and HIP5016 are a widely used pair of SynchroFETs from Harris Semiconductor. A 5 V to 3.3 V DC/DC converter using a generic PWM controller and the HIP5015 SynchroFET is shown in FIG. 22. This typically is used to derive 3.3 V from a 5 V input. The implementation of a two-output converter using these SynchroFET ICs is de scribed in Goodenough (1996). The HIP5015/5016 is designed to run at over 1 MHz. The efficiency of a 5 to 3.3 V converter running at 400 kHz using this IC is reported to be 85% for load currents between 0.5 and 4.5 A and over 90% for load currents between 0.8 and 2.8 A (Goodenough, 1996). The newer HIP5020 synchronous buck converter (Harris Corp., 1997) is optimized for battery-powered systems with 4.5-18 V input.

This buck converter can be operated up to a switching frequency of 1 MHz and demonstrates an efficiency of over 95%.

FIG. 21 A synchronous buck converter

FIG. 22 A PWM synchronous DC/DC converter using a SynchroFET. (Harris Semiconductor.)

4.5.2 Increased Gain and Bandwidth Control Loop

In low-voltage synchronous converters, the preferred method of control is the voltage mode. The power dissipated in the current sense resistor in current mode control cuts the efficiency by about 2% and therefore is unsuitable for low-voltage, high-efficiency applications. However, the low bandwidth of the voltage mode control loop reduces the converter's response to dynamic load conditions.

Maxim's three PWM controllers, the MAX796/797/799 designed for portable computer and communication applications, are current mode controllers that drive synchronous rectifiers, primarily in the buck mode. Efficiencies as high as 97% are reported by Maxim Integrated Circuits (1997). FIG. 23 shows the MAX797 controller. The special features of this circuit are a proprietary PWM comparator for handling transients, a proprietary idle mode control scheme used at low-load conditions for extended battery life, and the reduction of PWM noise.

This series is designed for output voltages as low as 2.5 V but can provide 1.5 V with external circuitry. To provide the required DC accuracy at the low-output voltage while handling high-speed current transients, these devices require an external op amp in the error amplifier circuit, which operates as an integrator (Maxim Integrated Circuits, 1997). The heart of this PWM controller is a multi-input open-loop comparator that sums the output voltage error with respect to the reference, the current sense signal, and a slope compensation ramp. This is of the direct summing type and lacks the traditional error amplifier with the associated phase shift. This direct summing configuration approaches the ideal of direct cycle-by-cycle control of the output voltage. This PWM comparator is shown in FIG. 24.

FIG. 23 A block diagram of the MAX797 PWM controller. (Maxim Integrated Circuits.)

FIG. 24 A block diagram of the main PWM comparator in the MAX797.

4.5.3 Idle Mode Control Scheme

The MAX797's approach for conservation of battery power uses the SKIP input shown in FIG. 23. At light loads (SKIP = 0) the inductor current fails to exceed the 30 mV threshold set by the minimum current comparator. When this occurs, the minimum current comparator immediately resets the high-side latch at the beginning of the cycle unless the output voltage drops below the reference.

This sends the controller into a variable-frequency idle mode, skipping most of the oscillator pulses, to cut back on gate-charge losses.

Operation at a fixed frequency, regardless of load conditions, however, is advantageous in terms of lowering PWM noise interference. Steps can be taken to remove the known emissions at fixed frequencies. Therefore, a low-noise mode is enabled in the MAX797 by making SKIP high. This forces fixed frequency operation by disabling the minimum current comparator.

4.5.4 Updated Voltage Mode Control

FIG. 25 A block diagram of the LTC 1430 controller. (by Linear Technology Corporation.)

A newer solution by Linear Technology employs an advanced voltage mode control scheme in its LTC1430 controller. This controller, optimized for high-power, 5 V/3.x V applications, operates at an efficiency greater than 90% in converter designs from 1 A to greater than 50 A output current (Linear Technology Corp., 1995). The LTC1430 uses a synchronous switching architecture with two N-channel output devices.

A block diagram of the LTC1430 is shown in FIG. 25. The primary control loop is a conventional voltage mode feedback loop. The load voltage is sensed across the output capacitor by the SENSE+ and SENSE-inputs, divided and fed to the feedback amplifier, where it’s compared to a 1.26 V internal reference. The error signal then is compared to a sawtooth waveform from the oscillator to generate a pulse width modulated switching waveform.

Two other comparators in the feedback loop, MIN and MAX, provide high-speed fault correction in situations where the feedback amplifier may not respond quickly enough. MIN compares the feedback signal to a voltage 40 mV (3%) below the reference, and MAX compares it to a voltage 3% above the reference.

If the output falls below the minimum level, the MIN comparator overrides the feedback comparator and forces the loop to full duty cycle (about 90%). If the output rises above the 3% level, the MAX comparator forces the duty cycle to 0. These two comparators prevent extreme output perturbations in the presence of fast output transients.

Additionally, the controller senses output current across the drain source resistance of the upper N-channel FET, providing an adjustable current limit without an external sense resistor.

4.5.5 A Master/Slave Architecture

Power Trends' solution (Travis, 1996) is a master/slave architecture for DC/DC converter modules, which allows the increase of available current in 3 A steps. This design consists of a 5 V/3.3 V converter with a 3 or 8 A master coupled with three 3 A slaves. The slaves add 9 A of current to the master's current output. This approach is an attractive solution, which leaves room for future system upgrades.

4.5.6 Improved Process Technologies

Aiming at improved control loop bandwidth, Siliconix developed a high-speed CBiC/D (complementary bipolar/CMOS/DMOS) process that provides reference. The error signal then is compared to a sawtooth waveform from the oscillator to generate a pulse width modulated switching waveform.

Two other comparators in the feedback loop, MIN and MAX, provide high-speed fault correction in situations where the feedback amplifier may not respond quickly enough. MIN compares the feedback signal to a voltage 40 mV (3%) below the reference, and MAX compares it to a voltage 3% above the reference.

If the output falls below the minimum level, the MIN comparator overrides the feedback comparator and forces the loop to full duty cycle (about 90%). If the output rises above the 3% level, the MAX comparator forces the duty cycle to 0. These two comparators prevent extreme output perturbations in the presence of fast output transients.

Additionally, the controller senses output current across the drain source resistance of the upper N-channel FET, providing an adjustable current limit without an external sense resistor.

FIG. 26 The principle of operation of switched capacitor converters

4.5.5 A Master/Slave Architecture

Power Trends' solution (Travis, 1996) is a master/slave architecture for DC/DC converter modules, which allows the increase of available current in 3 A steps. This design consists of a 5 V/3.3 V converter with a 3 or 8 A master coupled with three 3 A slaves. The slaves add 9 A of current to the master's current output. This approach is an attractive solution, which leaves room for future system upgrades.

FIG. 27 A simple IC switched capacitor implementation

4.5.6 Improved Process Technologies

Aiming at improved control loop bandwidth, Siliconix developed a high-speed CBiC/D (complementary bipolar/CMOS/DMOS) process that provides and small handheld communication devices is increasing as their output current capabilities are increasing and the supply current required by the portable devices is decreasing.

Simple switched capacitor converters such as the MAX660 can generate 100 mA at 3.3 V when powered from a two-cell battery of alkaline, NiCd, or NiMH cells or a single primary lithium cell. A disadvantage of a circuit such as this is the lack of regulation, which is overcome by adding a regulator externally.

Internal regulation in monolithic switched capacitor converter chips is achieved either as linear regulation or as charge pump modulation. Linear regulation offers low output noise. Charge pump modulation controls the switch resistance and offers more output current for a given cost or size because of the absence of a series-pass resistor. Newer charge pump ICs employ an on-demand switching technique that enables low quiescent current and high output current capability at the same time.

The simplicity of switched capacitor circuits, and hence its suitability for miniature equipment, is amply illustrated by Siliconix Inc. (1994b), where the eight-pin Si7660 is used for voltage inversion, doubling, and splitting.

4.5.8 SEPIC

The SEPIC (single-ended primary inductance converter) architecture is especially suited for battery-operated equipment and battery charger applications. The SEPIC topology is shown in FIG. 28.

FIG. 28 The SEPIC converter topology: (a) The basic circuit, (b) Associated waveforms

The two inductors L1 and L2 often are two identical windings on the same core. This topology essentially is similar to a 1:1 transformer flyback converter, except for the addition of a capacitor C, which forces identical AC voltages across the two inductors. This is a step-up/step-down topology with no inversion and no transformer. However, the SEPIC topology provides DC isolation from the input to the output through the presence of the capacitor.

The input/output relationship for this converter is given by

Vout --- VinD/(1 - D)

This topology has the advantage of being operable over a wide input voltage range. Due to this, the topology commonly finds applications in battery-powered equipme0t and battery chargers.

A converter design using the LT1373 current mode switching regulator with a SEPIC converter for generating 3.3 V from a single lithium-ion battery is described in Essaff (1995). In a typical application such as this, the battery voltage at full charge is above the output voltage; and when discharged, it’s below the output voltage. This presents special difficulties in the converter design. The SEPIC architecture, with its wide input voltage range, is ideally suited for such applications.

The use of the SEPIC topology in battery charging is illustrated in an application by Linear Technology Corp. (1996). The topology allows charging even when the input voltage is lower than the battery voltage. Further, it allows the current sense circuit to be ground-referenced and completely separated from the battery itself, simplifying battery switching and system grounding problems.

The LT1512 is a constant current controller for SEPIC converters, for charging NiCd and NiMH batteries. It also can provide a constant voltage source for charging lithium-ion batteries (Linear Technology Corp., 1996). A special feature of this controller is an internal low-dropout regulator, which provides a 2.3 V supply for all internal circuitry, allowing input voltages to vary from 2.7 to 25 V. This enables charging batteries from varied sources such as wall adapters, car batteries, and solar cells.

4.6 Illustrative DC to DC Converter ICs and Applications

The modularity of design, the wide-ranging power supply requirements, and the need for compactness, portability, and expandability of today's electronic equipment all contribute to the numerous applications that DC/DC converters have found in modern electronic equipment.

The area in which DC/DC converters have found the widest application is in distributed power systems. A distributed power system uses many small regulated power supplies, each located as close as possible to the load (Goodenough, 1995a). The bulk supply develops and distributes via a bus, an arbitrary voltage level. At appropriate points DC/DC converters change this voltage to the levels needed for the local circuitry. A distributed power system can save space and reduce the weight of the system. It also can improve reliability and the quality of the generated power and facilitate modular design and system expansion. Issues relating to distributed power systems are discussed in Ormond (1990, 1992) and Goodenough (1995a). A typical application of a distributed power system is in telecommunication equipment. A relatively high-voltage battery of 48 V is located in the bottom of each electronic equipment rack and DC/DC converters are used on each rack card cage to step down the voltage to the required level. Another such application is in aircraft power distribution systems.

Modern electronic systems contain both analog and digital circuitry, where analog components require voltage levels such as 9, -4-12, and 4-15. In mixed mode logic circuits, the conventional 5 V supply and a 3.3 V supply for the latest low-power ICs are required. In such complex cases, DC/DC converters perform the valuable function of generating all required voltages while saving cost and space. A common application of 5-12 V converters is in flash memory programmers, which require a supply of 12 V. Battery chargers that can be used off diverse power supplies, ranging from the conventional main supply to solar cells, find the step-up/step-down ability of DC/DC converters a useful feature.

Another application area of DC/DC converters is in small, battery-operated, portable equipment such as pagers, cameras, cellular phones, laptop computers, remote data acquisition, and instrumentation systems. The high efficiency and the small size of DC/DC converters become especially useful in such applications.

4.6.1 DC/DC Converter ICs

An increasing variety of chip-level DC/DC converters are appearing on the market. These devices not only are changing the way system designers structure their power supplies but are providing solutions to applications that previously required more costly, bulky, and cumbersome approaches.

Integrated circuit DC/DC converters are available in a wide range of power and other capabilities. The lower-power ICs can supply the exact voltage needed for a specific IC board, and higher-power types can simplify the design by reducing the component count (Pryce, 1988).

The key parameters and capabilities of modern low-power DC/DC converters are described in this section using a representative sample from principal manufacturers such as Maxim Integrated Circuits, Datel Inc., Linear Technology Corp., and Harris Semiconductor. For example, output power capabilities range from less than 1 W to more than 300 W. Single, double, or triple output configurations are available. Common output voltages are 3.3, 5, 12, and 15 V, while input voltages are quite varied. Common devices perform 5 to 12 V, 9 to 12 V, and 5 to 3.3 V conversions. TBL. 6 shows key features of some representative DC/DC converters.

Linear Technology Corp. produces a wide range of DC/DC converters, ranging from 0.5 W micro-power devices to high-efficiency 5 A devices. Simple voltage doublers and dividers as well as circuits that can be used in many configurations are available. The LT1073, a gated oscillator mode IC, can operate from a supply of 1 V and typically is used to generate 5 or 12 V from a single cell.

The XWR (wide input range) series by Datel Inc. (1991) is composed of high-efficiency current mode converters with typical efficiencies of about 85%. Common applications of these are in telecommunication, automotive, avionic, and marine equipment and in portable battery-operated systems. In addition to the XWR series, Datel also has the LP series of converters having 3, 4.5, 5, and 10 W outputs.

Maxim Integrated Circuits also developed a wide variety of DC/DC converters, operating in current mode and resonant mode. The MAX632 is a typical low-power device, which can operate from an input voltage of 5 V to produce a 12 V output. This type of converter is ideal for powering low-power analog circuits from a 5 V digital bus. These converters are relatively inexpensive and require very few external components.

TBL. 6 Some Representative DC/DC Converter ICs

The LM3578 from National Semiconductors is a low/medium power converter which can be used in buck, boost, and buck-boost configurations. This can supply output currents as high as 750 mA. Siliconix's Si9100 and Si9102 can handle high input voltages. These chips typically are used in transformer-coupled flyback and forward converter applications, such as ISDN and PABX equipment and modems. Due to their high input voltage ratings, the Si9100 can operate directly from the -48 V telephone line supply and the Si9102 from the -96 V double-battery telecom power supplies.

Harris Semiconductor's high-efficiency synchronous buck converter controller with integrated MOSFETs has typical applications in note book computers, portable instruments, and portable telecommunication equipment.

The switched capacitor converter ADP3603 from Analog Devices provides a -3 V regulated output from a 4.5-6.0 V supply and provides 50 mA of current. A 150 mA version, the ADP3604, also is available. Among the typical applications of these are negative voltage regulators, computer peripherals and add-on cards, pagers and radio control receivers, disk drives, and mobile phones.

Further details of these ICs as well as other similar products, application notes, and the like are found in Linear Technology Corp., Datel Inc., Siliconix, Motorola Inc., Unitrode Integrated Circuits, Malinaik, Travis, Good- enough, Sherman and Waiters, Maxim Integrated Circuits, Sherman, Pflasterer, "Trends in Battery Power", Williams and Huffman, Essaff, Harris Corp., and Analog Devices Inc..