chap20.pdf

Soft Switching

In addition to the resonant circuits introduced in Chapter 19‚ there has been much interest in reducing the

switching loss of the PWM converters of the previous chapters. Several of the more popular approaches

to obtaining soft switching in buck‚ boost‚ and other converters‚ are discussed in this chapter.

Mechanisms that cause switching loss are discussed in Chapter 4‚ including diode reverse

recovery‚ semiconductor output capacitances‚ and IGBT current tailing. Soft switching involves mitiga-

tion of one or more of these switching loss mechanisms in a PWM converter. The energy that would oth-

erwise be lost is recovered‚ and is transferred to the converter source or load. The operation of a

semiconductor device‚ during a given turn-on or turn-off switching transition‚ can be classified as hard-

switched‚ zero-current switched‚ or zero-voltage switched. Operation of diodes and transistors with soft

switching is examined in Section 20.1. In particular‚ it is preferable to operate diodes with zero-voltage

switching at their turn-off transitions‚ and to operate MOSFETs with zero-voltage switching during their

turn-on transitions. However‚ zero-voltage switching comes at the expense of increased conduction loss‚

and so the engineer must consider the effect of soft switching on the overall converter efficiency.

Resonant switch converters are a broad class of converters in which the PWM switch network of

a conventional buck‚ boost‚ or other converter is replaced with a switch cell containing resonant ele-

ments. These resonant elements are positioned such that the semiconductor devices operate with zero-

current or zero-voltage switching‚ and such that one or more of the switching loss mechanisms is reduced

or eliminated. Other soft-switching approaches may employ resonant switching transitions‚ but other-

wise exhibit the approximately rectangular waveforms of hard-switched converters. In any case‚ the

resulting hybrid converter combines the properties of the resonant switching network and the parent

hard-switched PWM converter.

Soft-switching converters can exhibit reduced switching loss‚ at the expense of increased con-

duction loss. Obtaining zero-voltage or zero-current switching requires that the resonant elements have

large ripple; often‚ these elements are operated in a manner similar to the discontinuous conduction

762

Soft Switching

modes of the series or parallel resonant converters. As in other resonant schemes‚ the objectives of

designing such a converter are: (1) to obtain smaller transformer and low-pass filter elements via increase

of the switching frequency‚ and/or (2) to reduce the switching loss induced by component nonidealities

such as diode stored charge‚ semiconductor device capacitances‚ and transformer leakage inductance and

winding capacitance.

The resonant switch and soft-switching ideas are quite general‚ and can be applied to a variety

of topologies and applications. A large number of resonant switch networks have been documented in the

literature; a few basic approaches are listed here [1–24]. The basic zero-current-switching quasi-resonant

switch network is analyzed in detail in Section 20.2. Expressions for the average components of the

switch network terminal waveforms are found‚ leading to determination of the switch conversion ratio

The switch conversion ratio performs the role of the duty cycle d of CCM PWM switch networks. For

example‚ the buck converter exhibits conversion ratio M equal to Both half-wave and full-wave ring-

ing of the tank network is considered; these lead to different switch conversion ratiofunctions In gen-

eral‚ given a PWM CCM converter having conversion ratio M ( d ) ‚ we can replace the PWM switch

network with a resonant switch network having switch conversion ratio The resulting quasi-resonant

converter will then have conversion ratio So we can obtain soft-switching versions of all of the

basic converters (buck‚ boost‚ buck-boost‚ forward‚ flyback‚ etc.)‚ that exhibit zero-voltage or zero-cur-

rent switching and other desirable properties.

In Section 20.3‚ the characteristics of several other resonant switch networks are listed: the

zero-voltage-switching quasi-resonant switch network‚ the zero-current-switching and zero-voltage-

switching quasi-square-wave networks‚ and the multiresonant switch network. One can obtain zero-volt-

age switching in all transistors and diodes using these networks.

Several related soft-switching approaches are now popular‚ which attain zero-voltage switching

of the transistor or transistors in commonly-used converters. The zero-voltage transition approach finds

application in full-bridge buck-derived converters. Active-clamp snubbers are often added to forward and

flyback converters‚ to attain zero-voltage switching and to reset the transformer. These circuits lead to

zero-voltage switching of the transistors‚ but (less-than-optimal) zero-current switching of the second-

ary-side diodes. Nonetheless‚ high efficiency can be achieved. An auxiliary resonant-commutated pole

can achieve zero-voltage switching in voltage-source inverters. These converters are briefly discussed in

Section 20.4.

20.1

SOFT-SWITCHING MECHANISMS

OF SEMICONDUCTOR DEVICES

When loosely used‚ the terms “zero-current switching” and “zero-voltage switching” normally refer to

one or more switching transitions of the transistor in a converter. However‚ to fully understand how a

converter generates switching loss‚ one must closely examine the switching transitions of every semicon-

ductor device. As described in Section 4.3‚ there are typically several mechanisms that are sources of sig-

nificant switching loss. At the turn-off transition of a diode‚ its reverse-recovery process can induce loss

in the transistor or other elements of the converter. The energy stored in the output capacitance of a

MOSFET can be lost when the MOSFET turns on. IGBTs can lose significant energy during their turn-

off transition‚ owing to the current-tailing phenomenon. The effects of zero-current switching and zero-

voltage switching on each of these devices isdiscussed in detail below.

20.1

Soft-Switching Mechanisms of Semiconductor Devices

763

20.1.1 Diode Switching

As discussed in Chapter 4‚ the reverse-recovery process usually leads to significant switching loss associ-

ated with the turn-off transition of diodes. This is often the largest single source of loss in a hard-

switched converter. Normally‚ negligible loss is associated with the turn-on transition of power diodes.

Three types of diode turn-off transition waveforms are commonly encountered in modern switching con-

verters: hard switching‚ zero-current switching‚ and zero-voltage switching.

Figure 20.1 illustrates a conventional hard-switched PWM buck converter. The diode voltage

and current waveforms v ( t ) and i ( t ) are also illustrated‚ with an exaggerated reverse recovery time. The

output inductor current ripple is small. The diode turns off when the transistor is turned on; the reverse

recovery process leads to a negative peak current of large amplitude. The diode must immediately sup-

port the full reverse voltage and hence both v ( t ) and i ( t ) must change with large slopes during reverse

recovery. As described in Section 4.3.2‚ hard switching of the diode induces energy loss

in the tran-

sistor‚ given approximately by

where is the diode recovered charge and is the reverse recovery time‚ both taken to be positive quan-

tities. The recovered charge is relatively large because the slope di/dt is large during the turn-off transi-

tion. The resonant circuit formed by the diode output capacitance and the diode package and other

wiring inductances leads to ringing at the end of the reverse recovery time.

Figure 20.2 illustrates zero-current switching at the turn-off transition of a diode. The converter

example is a quasi-resonant zero-voltage switching buck converter (see Section 20.3.1). The output

inductor current ripple is again small. However‚ tank inductor is now connected in series with the

diode. The resulting diode current waveform i ( t ) changes with a limited slope as shown. The diode

reverse-recovery process commences when i ( t ) passes through zero and becomes negative. The negative

i ( t ) actively removes stored charge from the diode; during this reverse recovery time‚ the diode remains

forward-biased. When the stored charge is removed‚ then the diode voltage must rapidly change to

As described in Section 4.3.3‚ energy is stored in inductor at the end of the reverse recovery time‚

given by

764

Soft Switching

The resonant circuit formed by and the diode output capacitance then cause this energy to be circu-

lated between and This energy is eventually dissipated by parasitic resistive elements in the circuit‚

and hence is lost. Since Eqs. (20.1) and (20.2) are similar in form‚ the switching losses induced by the

reverse-recovery processes of diodes operating with hard switching and with zero-current switching are

similar in magnitude. Zero-current switching may lead to somewhat lower loss because the reduced di/dt

leads to less recovered charge Zero-current switching of diodes also typically leads to increased peak

inverse diode voltage during the ringing of and because of the relatively large value of

When a diode operates with hard switching or zero-current switching‚ and when substantial

inductance is present in series with the diode‚ then significant ringing is observed in the diode voltage

waveform. A resonant circuit‚ comprised of the series inductance and the diode output capacitance‚ is

excited by the diode reverse recovery process‚ and the resulting ringing voltage can be of large enough

magnitude to lead to breakdown and failure of the diode. A common example is the diodes on the sec-

ondary side of a hard-switched transformer-isolated converter; the resonant circuit is then formed by the

transformer leakage inductance and the diode output capacitance. Other examples are the circuits of

Figs. 20.2 and 20.36‚ in which the series inductance is a discrete tank inductor.

A simple snubber circuit that is often used to protect the diode from excessive reverse voltage is

20.1

Soft-Switching Mechanisms of Semiconductor Devices

765

illustrated in Fig. 20.3. Resistor R damps the ringing of the resonant circuit. Capacitor C prevents the

off-state voltage of the diode from causing excessive power loss in R. Nonetheless‚ the energy consumed

by R per switching period is typically greater than Eqs. (20.1) or (20.2).

Figure 20.4 illustrates zero-voltage switching at the turn-off transition of a diode. The figure

illustrates the example of a zero-voltage switching quasi-square wave buck converter‚ discussed in Sec-

tion 20.3.3. The output inductor of this converter assumes the role of the tank inductor‚ and exhibits

large current ripple that causes the current to reverse polarity. While the diode conducts‚ its current

i ( t ) is equal to When becomes negative‚ the diode continues to conduct until its stored charge

has been removed. The diode then becomes reverse-biased‚ and flows through capacitor and the

diode output capacitance The diode voltage and current both change with limited slope in this type of

switching‚ and the loss induced by the diode reverse-recovery process is negligible because the wave-

forms are not significantly damped by parasitic resistances in the circuit‚ and because the peak currents

during reverse recovery are relatively low. The diode stored charge and diode output capacitance both

behave as an effective nonlinear capacitor that can be combined with (or replace) tank capacitor

Snubber circuits such as Fig. 20.3 are not necessary when the diode operates with zero-voltage switch-

ing.

Thus‚ zero-voltage switching at the turn-off transition of a diode is the preferred approach‚ that

leads to minimum switching loss. Zero-current switching at the turn-off transition can be problematic‚

because of the high peak inverse voltage induced across the diode by ringing.

20.1.2 MOSFET Switching

The switching loss mechanisms typically encountered by a MOSFET in a hard-switched converter are

discussed in Chapter 4‚ and typical MOSFET voltage and current waveforms are illustrated in Fig. 20.5.

The most significant components of switching loss in the MOSFET of this circuit are: (1) the loss

Plik z chomika:

Inne pliki z tego folderu:

Inne foldery tego chomika: