10

Input Filter Design

10.1 10.1.1

INTRODUCTION Conducted EMI

It is nearly always required that a filter be added at the power input of a switching converter. By attenuating the switching harmonics that are present in the converter i nput current waveform, the i nput filter allows compliance w i th regulations that limit conducted electromagnetic interference (EMI). The i nput filter can also protect the converter and its load from transients that appear in the i nput voltage thereby i mprov i ng the system rel i ab i l i ty . A simple buck converter example is illustrated in Fig. 10.1. The converter injects the pulsat i ng current of Fig. 10.1(b) i nto the power source The Fourier series of contains harmonics at multiples of the switching frequency as follows:

378

Input Filter Design

In practice, the magn i tudes of the h i gher-order harmon i cs can also be s i gn i f i cantly affected by the current spike caused by diode reverse recovery, and also by the finite slopes of the switching transitions. The large high-frequency current harmonics of can interfere w i th television and radio reception, and can d i srupt the operation of nearby electron i c equ i pment . In consequence, regulat i ons and standards ex i st that l i m i t the ampl i tudes of the harmon i c currents injected by a sw i tch i ng converter i nto i ts power source [1-8]. As an example, if the dc inductor current i of Fig. 10.2 has a magnitude of several Amperes, then the fundamental component (n = 1) has an rms ampl i tude i n the v i c i n i ty of one Ampere. Regulat i ons may require attenuation of this current to a value typically in the range to To meet l i m i ts on conducted EMI, it is necessary to add an i nput f i lter to the converter. Figure 10.2 illustrates a simple single-section L-C low-pass filter, added to the i nput of the converter of Fig. 10.1. Th i s f i lter attenuates the current harmon i cs produced by the sw i tch i ng converter, and thereby smooths the current waveform drawn from the power source. If the filter has transfer function then the i nput current Fourier series becomes

In other words, the ampl i tude of each current harmon i c at angularfrequency is attenuated by the f i lter transfer funct i on at the harmon i c frequency, Typ i cal requ i rements effect i vely l i m i t the current harmonics to have amplitudes less than and hence i nput filters are often required to attenuate the current ampl i tudes by 80 dB or more. To improve the rel i ab i l i ty of the system, i nput f i lters are somet i mes requ i red to operate normally when trans i ents or periodic d i sturbances are applied to the power i nput . Such conducted susceptibility spec i f i cat i ons force the des i gner to damp the i nput f i lter resonances, so that i nput d i sturbances do not exc i te excessive currents or voltages w i th i n the f i lter or converter .

10.1 Introduction

379

10.1.2

The Input Filter Design Problem

The situation faced by the design engineer is typically as follows. A switching regulator has been designed, wh i ch meets performance spec i f i cat i ons . The regulator was properly designed as discussed in Chapter 9, us i ng a small-s i gnal model of the converter power stage such as the equ i valent circu i t of Fig. 10.3(a). In consequence, the trans i ent response is well damped and suff i c i ently fast, w i th adequate phase marg i n at all expected operat i ng po i nts . The output impedance is suff i c i ently small over a wide frequency range. The l i ne-to-output transfer funct i on or audiosusceptibility, is suff i c i ently small, so that the output voltage remains regulated in spite of var i at i ons in Hav i ng developed a good des i gn that meets the above goals regarding dynam i c response, the problem of conducted EMI is then addressed. A low-pass filter having attenuation sufficient to meet conducted EMI specifications is constructed and added to the converter i nput . A new problem then arises: the input filter changes the dynamics of the converter. The transient response is modified, and the control system may even become unstable . The output i mpedance may become large over some frequency range, possibly exh i b i t i ng resonances. The aud i osuscept i b i l i ty may be degraded. The problem is that the i nput filter affects the dynamics of the converter, often in a manner that degrades regulator performance . For example, when a s i ngle-sect i on L-C i nput filter is added to a buck converter as in Fig. 10.2(a), the small-signal equivalent circuit model is modified as shown in Fig. 10.3(b). The input filter elements affect all transfer functions of the converter, including the control-to-

380

Input Filter Design

output transfer funct i on the line-to-output transfer function and the converter output i mpedance Moreover, the i nfluence of the i nput f i lter on these transfer funct i ons can be qu i te severe. As an illustration, lets examine how the control-to-output transfer funct i on of the buck converter of Fig. 10.1 is altered when a s i mple L-C i nput f i lter is added as in Fig. 10.2. For th i s example, the element values are chosen to be: Figure 10.4 conta i ns the Bode plot of the magn i tude and phase of the control-to-output transfer funct i on The dashed l i nes are the magn i tude and phase before the i nput f i lter was added, generated by solut i on of the model of Fig. 10.3(a). The complex poles of the converter output f i lter cause the phase to approach 180 at high frequency. Usually, this is the model used to design the regulator feedabck loop and to evaluate the phase marg i n (see Chapter 9). The solid l i nes of Fig. 10.4 show the magn i tude and phase after add i t i on of the i nput filter, generated by solut i on of the model of Fig. 10.3(b). The magn i tude exh i b i ts a gl i tch at the resonant frequency of the i nput f i lter, and an add i t i onal - 360 of phase sh i ft is i ntroduced i nto the phase. It can be shown that now conta i ns an add i t i onal complex pole pa i r and a complex r i ght half-plane zero pair, associated w i th the i nput f i lter dynam i cs . If the crossover frequency of the regulator feedback loop is near to or greater than the resonant frequency of the i nput filter, then the loop phase marg i n w i ll become negat i ve and i nstab i l i ty w i ll result . Such behav i or i s typ i cal ; consequently, i nput f i lters are notor i ous for destab i l i z i ng sw i tch i ng regulator systems. This chapter shows how to mitigate the stability problem, by i ntroduc i ng damp i ng i nto the i nput f i lter and by des i gn i ng the i nput f i lter such that its output i mpedance i s suff i c i ently small [9-21]. The result of these measures i s that the effect of the i nput f i lter on the control-to-output transfer funct i on becomes negl i g i ble, and hence the converter dynam i cs are much better behaved . Although analys i s of the fourth-order system of Fig. 10.3(b) is potent i ally qu i te complex, the approach used here s i mpl i f i es the problem through use of i mpedance i nequal i t i es i nvolv i ng the converter i nput i mpedance and the f i lter output i mpedance [9,10]. These i nequal i t i es are based on M i ddlebrook s extra element theorem of Appendix C. This approach allows the engineer to gain the i ns i ght needed to effectively design the i nput filter. Optimization of the damping networks of input filters, and design of multiple-section filters, is also discussed.

10.2

Effect of an Input Filler on Converter Transfer Functions

381

10.2

EFFECT OF AN INPUT FILTER ON CONVERTER TRANSFER FUNCTIONS is defined as follows:

The control-to-output transfer funct i on

The control-to-output transfer functions of basic CCM converters w i th no i nput filters are listed in Section 8.2.2. Addition of an input filter to a switching regulator leads to the system illustrated in Fig. 10.5. To determine the control-to-output transfer function in the presence of the input filter, we set to zero and solve for according to Eq. (10.3). The i nput filter can then be represented s i mply by its output impedance as illustrated in Fig. 10.6. Thus, the input filter can be treated as an extra element having impedance In Appendix C, Section C.4.3, Middlebrooks extra element theorem is employed to determine how addition of the input filter modifies the control-to-output transfer function. It is found that the modified control-to-output transfer funct i on can be expressed as follows [9]:

382

Input Filter Design

where

is the original control-to-output transfer funct i on w i th no i nput filter. The quant i ty converter input impedance under the cond i t i on that is equal to zero:

is equal to the

The quant i ty is equal to the converter i nput impedance under the cond i t i on that the feedback controller of Fig. 10.5 operates ideally; in other words, the controller varies as necessary to ma i nta i n equal to zero:

In terms of the canon i cal c i rcu i t model parameters described in Section 7.5,

can be shown to be

Expressions for 10.1. 10.2.1 Discussion

and

for the basic buck, boost, and buck-boost converters are listed in Table

Equation (10.4) relates the power stage control-to-output transfer funct i on to the output impedance of the input filter, and also to the quantities and measured at the power input port of the converter. The quant i ty coincides w i th the open-loop i nput impedance of the converter. As described above, the quant i ty is equal to the i nput port i mpedance of the converter

10.2 Effect of an Input Filter on Converter Transfer Functions

383

power stage, under the cond i t i ons that is varied as necessary to null to zero. This is, in fact, the funct i on performed by an ideal controller: it varies the duty cycle as necessary to ma i nta i n zero error of the output voltage. Therefore, coincides w i th the impedance that would be measured at the converter power i nput term i nals, if an ideal feedback loop perfectly regulated the converter output voltage. Of course, Eq. (10.4) is valid in general, regardless of whether a control system is present. Figure 10.7 i llustrates the large-signal behavior of a feedback loop that perfectly regulates the converter output voltage. Regardless of the applied i nput voltage the output voltage is ma i nta i ned equal to the desired value V. The load power is therefore constant, and equal to In the idealized case of a lossless converter, the power flow i ng i nto the converter i nput term i nals w i ll also be equal to regardless of the value of Hence, the power i nput term i nal of the converter obeys the equation

This characteristic is i llustrated in Fig. 10.7(b), and is represented in Fig. 10.7(a) by the dependent power s i nk symbol. The properties of power sources and power s i nks are discussed in detail in Chapter 11. Figure 10.7(b) also illustrates linearization of the constant input power characteristic, about a quiescent operating po i nt . The result i ng l i ne has negative slope; therefore, the i ncremental (small signal) i nput resistance of the ideal voltage regulator is negative. For example, increasing the voltage

384

Input Filter Design

causes the current has the value [9,14]:

to decrease, such that the power rema i ns constant . Th i s i ncremental resistance

where R is the output load resistance, and M is the conversion rat i o For each of the converters listed in Table 10.1, the dc asymptote of coincides w i th the negative i ncremental resistance g i ven by the equation above. Practical control systems exhibit a limited bandw i dth, determined by the crossover frequency of the feedback loop. Therefore, we would expect the closed-loop regulator i nput i mpedance to be approximately equal to at low frequency where the loop gain is large and the regulator works well. At frequencies above the bandwidth of the regulator we expect the converter i nput impedance to follow the open-loop value For closed-loop cond i t i ons, it can be shown that the regulator i nput impedance is, in fact, described by the follow i ng equat i on :

where T(s) is the controller loop gain. Thus, the regulator input impedance follows the negative resistance of at low frequency where the magn i tude of the loop ga i n is large [and hence and reverts to the (positive) open-loop impedance at h i gh frequency where is small [i.e., where When an undamped or l i ghtly damped i nput filter is connected to the regulator i nput port, the i nput f i lter can interact w i th the negative resistance characteristic of to form a negative resistance oscillator. This further expla i ns why add i t i on of an i nput f i lter tends to lead to i nstab i l i t i es . 10.2.2 Impedance Inequalities

Equation (10.4) reveals that addition of the i nput filter causes the control-to-output transfer funct i on to be modified by the factor

called the correction factor. When the following inequalities are satisfied,

then the correction factor has a magn i tude of approxi mately un i ty, and the i nput filter does not substant i ally alter the control-to-output transfer funct i on [9,10]. These i nequal i t i es l i m i t the max i mum allowable output impedance of the i nput filter, and const i tute useful filter des i gn cr i ter i a . One can sketch the Bode and plots of and compare with the Bode plot of This allows the engineer to gain the i ns i ght necessary to design an i nput filter that satisfies Eq. (10.13).

10.3 Buck Converter Example

385

A s i m i lar analys i s shows that the converter output i mpedance is not substant i ally affected by the input filter when the following inequalities are satisfied:

where is again as given in Table 10.1. The quant i ty is equal to the converter input impedance under the cond i t i ons that the converter output is shorted:

Expressions for for basic converters are also listed in Table 10.1. S i m i lar i mpedance i nequal i t i es can be derived for the case of current-programmed converters [12,13], or converters operating in the discontinuous conduction mode. In [12], impedance inequalities nearly identical to the above equations were shown to guarantee that the i nput filter does not degrade transient response and stability in the current-programmed case. Feedforward of the converter input voltage was suggested in [16].

10.3

BUCK CONVERTER EXAMPLE

Let us again consider the example of a simple buck converter with L-C input filter, as illustrated in Fig. 10.8(a). Upon replacing the converter w i th its small-signal model, we obtain the equivalent circuit of Fig. 10.8(b). Lets evaluate Eq. (10.4) for this example, to find how the input filter modifies the control-tooutput transfer function of the converter. 10.3.1 Effect of Undamped Input Filter

The quant i t i es and can be read from Table 10.1, or can be derived us i ng Eqs. (10.6) and (10.7) as further described in Appendix C. Thequant i ty is given by Eq. (10.6). Upon setting to zero, the converter small signal model reduces to the c i rcu i t of Fig. 10.9(a). It can be seen that is equal to the input impedance of the R-L-C filter, divided by the square of the turns ratio:

Construction of asymptotes for th i s impedance is treated in Section 8.4, w i th the results for the numer i cal values of this example given in Fig. 10.10. The load resistance dominates the impedance at low frequency, lead i ng to a dc asymptote of For the high-Q case shown, follows the output capacitor asymptote, reflected through the square of the effective turns ratio, at intermediate frequencies. A series resonance occurs at the output filter resonant frequency given by

For the element values listed in Fig. 10.8(a), the resonant frequency is The values of the asymptotes at the resonant frequency are g i ven by the characteristic impedance referred to the

386

Input Filter Design

10.3

Buck Converter Example

387

transformer primary:

For the element values given in Fig. 10.8(a), this expression is equal to

The

factor is given by

This expression y i elds a numer i cal value of The value of at the resonant frequency 1.6 kHz i s therefore equal to At h i gh frequency, follows the reflected inductor asymptote. The quant i ty is given by Eq. (10.7). This impedance is equal to the converter input impedance under the cond i t i ons that is varied to ma i nta i n the output voltage at zero. Figure 10.9(b) illustrates the derivation of an expression for A test current source is injected at the to converter i nput port. The impedance can be viewed as the transfer funct i on from

The null condition greatly s i mpl i f i es analys i s of the c i rcu i t of Fig. 10.9(b). Since the voltage is zero, the currents through the capacitor and load impedances are also zero. This further implies that the inductor current and transformer w i nd i ng currents are zero, and hence the voltage across the inductor is also zero. Finally, the voltage equal to the output voltage plus the inductor voltage, is zero. Since the currents in the w i nd i ngs of the transformer model are zero, the current is equal to the independent source current

Because is equal to zero, the voltage applied to the secondary of the transformer model is equal to the i ndependent source voltage Upon d i v i d i ng by the turns rat i o D, we obtain

388

Input Filter Design

Insertion of Eqs. (10.21) and (10.22) i nto Eq. (10.20) leads to the follow i ng result:

The steady-state relat i onsh i p has been used to s i mpl i fy the above result. This equation coincides w i th the expression listed in Table 10.1. The Bode diagram of is constructed in Fig. 10.10; this plot coincides with the dc asymptote of Next, let us construct the Bode d i agram of the f i lter output i mpedance When the i ndependent source is set to zero, the i nput f i lter network reduces to the c i rcu i t of Fig. 10.11. It can be seen that is given by the parallel comb i nat i on of the inductor and the capacitor

Construction of the Bode diagram of th i s parallel resonant c i rcu i t is discussed in Section 8.3.4. As i llustrated in Fig. 10.12, the magn i tude is dominated by the inductor impedance at low frequency, and by the capacitor impedance at h i gh frequency. The i nductor and capacitor asymptotes intersect at the f i lter resonant frequency:

10.3 Buck Converter Example

389

For the given values, the input filter resonant frequency is impedance

Hz. This filter has characteristic

equal to Since the input filter is undamped, its Q- factor is ideally infinite. In practice, parasitic elements such as inductor loss and capacitor equ i valent series resistance l i m i t the value of Nonetheless, the impedance is very large in the vicinity of the filter resonant frequency The Bode plot of the filter output impedance is overlaid on the and plots in Fig. 10.13, for the element values listed in Fig. 10.8(a). We can now determine whether the impedance inequalities (10.13) are satisfied. Note the design-oriented nature of Fig. 10.13: since analytical expressions are given for each impedance asymptote, the designer can easily adjust the component values to satisfy Eq. (10.13). For example, the values of and should be chosen to ensure that the asymptotes of lie below the worst-case value of R/D2, as well as the other asymptotes of It should also be apparent that it is a bad idea to choose the input and output filter resonant frequencies and to be equal, because it would then be more d i ff i cult to satisfy the i nequal i t i es of Eq. (10.13). Instead, the resonant frequencies and should be well separated in value. Since the i nput filter is undamped, it is impossible to satisfy the impedance inequalities (10.13) in the vicinity of the input filter resonant frequency Regardless of the choice of element values, the input filter changes the control-to-output transfer function in the vicinity of frequency Figures 10.14 and 10.15 illustrate the resulting correction factor [Eq. (10.12)] and the modified control-to-output transfer function [Eq. (10.4)], respectively. At frequencies well below the i nput filter resonant frequency, impedance inequalities (10.13) are well satisfied. The correction factor tends to the value and the

390

Input Filter Design

control-to-output transfer funct i on i s essent i ally unchanged . In the v i c i n i ty of the resonant frequency the correction factor conta i ns a pa i r of complex poles, and also a pa i r of r i ght half-plane complex zeroes. These cause a gl i tch in the magn i tude plot of the correction factor, and they contr i bute 360 of lag to the phase of the correction factor. The gl i tch and its phase lag can be seen in the Bode plot of At high frequency, the correction factor tends to a value of approximately consequently, the h i gh-frequency magn i tude of is unchanged. However, when the 360 contributed by the correction factor is added to the 180 contributed at high frequency by the two poles of the original a high-frequency phase asymptote of 540 is obtained. If the crossover frequency of the converter feedback loop i s placed near to or greater than the i nput f i lter resonant frequency then a negative

10.3 Buck Converter Example

391

phase margin is inevitable. This expla i ns why addition of an i nput f i lter often leads to instabilities and oscillations in switching regulators. 10.3.2 Damping the Input Filter

Lets damp the resonance of the input filter, so that impedance inequalities (10.13) are satisfied at all frequencies. One approach to damping the filter is to add resistor in parallel w i th capacitor as illustrated in Fig. 10.16(a). The output i mpedance of th i s network is identical to the parallel resonant impedance analyzed in Section 8.3.4. The max i mum value of the output i mpedance occurs at the resonant frequency and is equal in value to the resistance Hence, to satisfy impedance inequalities (10.13), we should and choose to be much less than the asymptotes. The condition can be expressed as:

Unfortunately, th i s raises a new problem: the power dissipation in The dc i nput voltage is appl i ed across resistor and therefore dissipates power equal to Equat i on (10.27) i mpl i es that th i s power loss is greater than the load power! Therefore, the circuit of Fig. 10.16(a) is not a practical solution. One solut i on to the power loss problem is to place in parallel w i th as i llustrated in Fig. 10.16(b). The value of in Fig. 10.16(b) is also chosen according to Eq. (10.27). Since the dc voltage across i nductor is zero, there is now no dc power loss in resistor The problem w i th th i s c i rcu i t is that its transfer funct i on conta i ns a h i gh-frequency zero. Add i t i on of degrades the slope of the h i ghfrequency asymptote, from 40 dB/decade to 20 dB/decade. The c i rcu i t of Fig. 10.16(b) is effectively a single-pole R-C low-pass filter, w i th no attenuation provided by i nductor One practical solution is i llustrated in Fig. 10.17 [10]. Dc block i ng capacitor is added in series w i th resistor Since no dc current can flow through resistor its dc power loss is el i m i nated . The value of is chosen to be very large such that, at the filter resonant frequency the impedance of the branch is dom i nated by resistor When is suff i c i ently large, then the output i mpedance of th i s network reduces to the output impedances of the filters of Fig. 10.16. The impedance asymptotes for the case of large are illustrated in Fig. 10.17(b).

392

Input Filter Design

The low-frequency asymptotes of and in F i g . 10.10 are equal to The choice therefore sat i sf i es i mpedance i nequal i t i es (10.13) very well . The choice leads to which is much smaller than The result i ng magn itude is compared w i th and in Fig. 10.18. It can be seen that the chosen values of and lead to adequate damp i ng, and impedance i nequal i t i es (10.13) are now well sat i sf i ed . Figure 10.19 i llustrates how add i t i on of the damped i nput f i lter mod i f i es the magn i tude and phase of the control-to-output transfer funct i on . There is now very l i ttle change in and we would expect that the performance of the converter feedback loop is unaffected by the i nput filter.

10.4

DESIGN OF A DAMPED INPUT FILTER

As i llustrated by the example of the prev i ous sect i on, des i gn of an i nput f i lter requ i res not only that the f i lter i mpedance asymptotes sat i sfy i mpedance i nequal i t i es, but also that the filter be adequately damped. Damp i ng of the i nput f i lter is also necessary to prevent trans i ents and d i sturbances in from exc i t i ng filter resonances. Other design constraints include attaining the desired filter attenuation, and m i n i m i z i ng

10.4

Design of a Damped Input Filter

393

the size of the reactive elements. Although a large number of classical f i lter design techn i ques are well known, these techniques do not address the problems of l i m i t i ng the max i mum output impedance and damp i ng f i lter resonances. The value of the block i ng capacitor used to damp the i nput filter in Section 10.3.2 is ten times larger than the value of and hence its size and cost are of practical concern. Optimization of an input filter design therefore includes minimization of the size of the elements used in the damping networks. Several practical approaches to damp i ng the single-section L-C low-pass filter are i llustrated in Fig. 10.20 [10,11,17]. Figure 10.20(a) contains the damping branch considered in the previous section. In Fig. 10.20(b), the damping resistor is placed in parallel w i th the filter inductor and a high-frequency blocki ng inductor is placed in series w i th Inductor causes the filter transfer function to roll off with a high-frequency slope of 40 dB/decade. In Fig. 10.20(c), the damping resistor is placed in series w i th the filter inductor and the dc current is bypassed by inductor In each case, it is desired to obtain a given amount of damp i ng [i.e., to cause the peak value of the filter output impedance to be no greater than a given value that sat i sf i es the i mpedance i nequal i t i es (10.13)], wh i le minimizing the value of or . This problem can be formulated in an alternate but equivalent form: for a given choice of or findthe value of that m i n i m i zes the peak output impedance [10]. The solutions to this opt i m i zat i on problem, for the three filter networks of Fig. 21, are summar i zed in this section. In each case, the quantities and are defined by Eqs. (10.25) and (10.26). Consider the filter of Fig. 10.20(b), with fixed values of and Figure 10.21 contains Bode plots of the filter output impedance for several values of damp i ng resistance For the limiting case the circuit reduces to the original undamped filter with infinite In the l i m i t i ng case the filter is also undamped, but the resonant frequency is increased because becomes connected in parallel with Between these two extremes, there must exist an optimum value of that causes the peak filter output impedance to be m i n i m i zed . It can be shown [10,17] that all magn i tude plots must pass through a common point, and therefore the opt i mum atta i ns i ts peak at th i s po i nt . This fact has been used to derive the design equations of optimally-damped L-C filter sections.

394

Input Filter Design

10.4

Design of a Damped Input Filter

395

10.4.1

Parallel Damping

Optimization of the filter network of Fig. 10.20(a) and Section 10.3.2 was described in [10]. The highfrequency attenuation of this filter is not affected by the choice of and the h i gh-frequency asymptote is identical to that of the original undamped filter. The sole tradeoff in design of the damping elements for this filter is in the size of the blocking capacitor vs. the damp i ng achieved. For th i s filter, let us def i ne the quant i ty as the ratio of the blocking capacitance to the filter capacitance

For the opt i mum design, the peak filter output impedance occurs at the frequency

The value of the peak output impedance for the opt i mum design is

The value of damping resistance that leads to optimum damping is described by

The above equations allow choice of the damp i ng values and For example, lets redesign the damping network of Section 10.3.2, to achieve the same peak output impedance wh i le m i n i m i z i ng the value of the blocking capacitance From Section 10.3.2, the other parameter values are and First, we solve Eq. (10.30) to find the required value of

Evaluation of this expression w i th the given numerical values leads to The block i ng capacitor is therefore required to have a value of This is one-quarter of the value employed in Section 10.3.2. The value of is then found by evaluation of Eq. (10.31), leading to

The output impedance of this filter design is compared w i th the output impedances of the original undamped filter of Section 10.3.1, and of the suboptimal design of Section 10.3.2, in Fig. 10.22. It can be

396

Input Filter Design

seen that the opt i mally damped filter does indeed achieve the desired peak output impedance of at the slightly lower peak frequency given by Eq. (10.29) The parallel damping approach finds significant application in dcdc converters. Since a series resistor is placed in series w i th can be realized us i ng capacitor types hav i ng substant i al equivalent series resistance, such as electrolytic and tantalum types. However, in some appl i cat i ons, the approaches of the next subsections can lead to smaller designs. Also, the large block i ng capacitor value may be undesirable in applications hav i ng an ac i nput . 10.4.2 Parallel Damping

Figure 10.20(b) illustrates the placement of damp i ng resistor in parallel w i th inductor Inductor causes the filter to exh i b i t a two-pole attenuat i on characteri st i c at h i gh frequency. To allow to damp the filter, inductor should have an impedance magn i tude that is suff i c i ently smaller than at the filter resonant frequency Optimization of th i s damp i ng network is described in [17]. With th i s approach, i nductor can be phys i cally much smaller than Since is typically much greater than the dc resistance of essent i ally none of the dc current flows through Furthermore, could be realized as the equivalent series resistance of at the filter resonant frequency Hence, th i s is a very simple, low-cost approach to damp i ng the i nput filter. The disadvantage of this approach is the fact that the h i gh-frequency attenuat i on of the filter is degraded: the h i gh-frequency asymptote of the filter transfer funct i on is increased from to Furthermore, since the need for damping l i m i ts the max i mum value of s i gn i f i cantloss of h i gh-frequency attenuation is unavo i dable . To compensate, the value of must be increased. Therefore, a tradeoff occurs between damping and degradation of h i gh-frequency attenuat i on, as i llustrated in Fig. 10.23. For example, l i m i t i ng the degradation of h i gh-frequency attenuat i on to 6 dB leads to an opt iof mum peak filter output impedance times the or i g i nal character i st i c impedance Add i tional damping leads to further degradation of the h i gh-frequency attenuat i on . The optimally damped design (i.e., the choice of that m i n i m i zes the peak output impedance

10.4 Design of a Damped Input Filter

397

for a given choice of

is described by the follow i ng equations:

where

The peak filter output impedance occurs at frequency

and has the value

The attenuation of the filter high-frequency asymptote is degraded by the factor

So, given an undamped filter having corner frequency and characteristic impedance and given a requirement for the max i mum allowable output impedance one can solve Eq. (10.37) for the required value of One can then determi ne the required numer i cal values of and

398

Input Filter Design

10.4.3

Series Damping

Figure 10.20(c) i llustrates the placement of damp i ng resistor in series w i th i nductor Inductor provides a dc bypass to avoid s i gn i f i cant power d i ss i pat i on in To allow to damp the filter, i nductor should have an impedance magn i tude that i s suff i c i ently greater than at the f i lter resonant frequency . Although th i s c i rcu i t is theoret i cally equ i valent to the parallel damp i ng case of Section 10.4.2, several differences are observed in practical designs. Both i nductors must carry the full dc current, and hence both have s i gn i f i cant size. The f i lter h i gh-frequency attenuat i on is not affected by the choice of and the h i gh-frequency asymptote is i dent i cal to that of the or i g i nal undamped filter. The tradeoff in design of this filter does not i nvolve h i gh-frequency attenuat i on ; rather, the issue is damp i ng vs. bypass i nductor size. Design equat i ons s i m i lar to those of the prev ious sections can be derived for th i s case. The opt i mum peak filter output impedance occurs at frequency

and has the value

The value of damp i ng resistance that leads to opt i mum damp i ng is described by

For th i s case, the peak output i mpedance cannot be reduced below theless, it is possible to further reduce the filter output impedance by redesign of value of 10.4.4 Cascading Filter Sections

v i a damp i ng . Noneand to reduce the

A cascade connection of mult i ple L-C filter sections can achieve a given h i gh-frequency attenuat i on w i th less volume and we i ght than a single-section L-C filter. The increased cutoff frequency of the mult i plesection filter allows use of smaller i nductance and capacitance values. Damp i ng of each L-C section is usually required, wh i ch implies that damp i ng of each section should be opt i m i zed . Unfortunately, the results of the previous sections are restricted to single-section filters. Interact i ons between cascaded L-C sections can lead to add i t i onal resonances and increased filter output i mpedance . It is nonetheless possible to design cascaded filter sections such that i nteract i on between L-C sections is negl i g i ble . In the approach described below, the filter output i mpedance is approx i mately equal to the output impedance of the last sect ion, and resonances caused by i nteract i ons between stages are avoided. Although the result i ng filter may not be opt i mal in any sense, i ns i ght can be gained that allows i ntell i gent design of mult i ple-sect i on filters w i th econom i cal damp i ng of each section.

10.4

Design of a Damped Input Filter

399

Consider the addition of a filter section to the i nput of an ex i st i ng filter, as in Fig. 10.24. Let us assume that the ex i st i ng f i lter has been correctly designed to meet the output i mpedance design cr i ter i a of Eq. (10.13): under the cond i t i ons is suff i c i ently small . It is desired to add a and damped filter section that does not s i gn i f i cantly increase Middlebrooks extra element theorem of Append i x C can again be invoked, to express how addition of the filter section modifies

where

is the impedance at the i nput port of the ex i st i ng filter, w i th its output port short-circuited. Note that, in this particular case, null i ng is the same as shorting the filter output port because the short-c i rcu i t current flows through the source. The quant i ty

is the impedance at the input port of the existing filter, with its output port open-circuited. Hence, the additional filter section does not s i gn i f i cantly alter provided that

Bode plots of the quant i t i es and can be constructed e i ther analyt i cally or by computer s i mulat i on, to obtain l i m i ts of When sat i sf i es Eq. (10.45), then the correction factor is approx i mately equal to 1, and the modified is approx i mately equal to the or i g i nal To sat i sfy the design cr i ter i a (10.45), it is advantageous to select the resonant frequencies of to differ from the resonant frequencies of In other words, we should stagger-tune the filter sections. This minimizes the interactions between filter sections, and can allow use of smaller reactive element values.

400

Input Filter Design

10.4.5

Example: Two Stage Input Filter

As an example, let us consider the design of a two-stage filter using parallel damping in each section as illustrated in Fig. 10.25 [17]. It is desired to achieve the same attenuat i on as the single-section f i lters designed in Sections 10.3.2 and 10.4.1, and to f i lter the i nput current of the same buck converter example of Fig. 10.8. These filters exh i b i t an attenuat i on of 80 dB at 250 kHz, and sat i sfy the design inequal i t i es of Eq. (10.13) w i th the and impedances of Fig. 10.10. Hence, lets design the filter of Fig. 10.25 to attain 80 dB of attenuation at 250 kHz. As described in the previous section and below, it is advantageous to stagger-tune the filter sections so that interaction between filter sections is reduced. We w i ll find that the cutoff frequency of f i lter section 1 should be chosen to be smaller than the cutoff frequency of section 2. In consequence, the attenuation of section 1 w i ll be greater than that of section 2. Let us (somewhat arb i trar i ly) design to obtain 45 dB of attenuat i on from section 1, and 35 dB of attenuation from section 2 (so that the total is the specified 80 dB). Let us also select for each section; as i llustrated in Fig. 10.23, this choice leads to a good compromise between damp i ng of the f i lter resonance and degradation of h i gh frequency filter attenuat i on . Equation (10.38) and Fig. 10.23 predict that the damp i ng network w i ll degrade the h i gh frequency attenuat i on by a factor of or 9.5 dB. Hence, the section 1 undamped resonant frequency should be chosen to y i eld of attenuation at 250 kHz. Since section 1 exhibits a two-pole ( 40 dB/decade) roll-off at high frequencies, should be chosen as follows:

Note that th i s frequency is well above the 1.6 kHz resonant frequency of the buck converter output filter. Consequently, the output impedance can be as large as and st i ll be well below the plots of Fig. 10.10. and Solution of Eq. (10.37) for the required section 1 character i st i c impedance that leads to a peak output impedance of with n = 0.5 leads to

10.4

Design of a Damped Input Filter

401

The filter inductance and capacitance values are therefore

The section 1 damping network inductance is

The section 1 damping resistance is found from Eq. (10.34):

The peak output impedance will occur at the frequency given by Eq. (10.36), 15.3 kHz. The quantities and for filter section 1 can now be constructed analytically or plotted by computer simulation. is the section 1 input impedance with the output of section 1 shorted, and is given by the parallel combination of the and the branches. is the section 1 input impedance with the output of section 1 open-circuited, and is given by the series combination of with the capacitor impedance Figure 10.26 contains plots of for and filter section 1, generated using Spice. One way to approach design of filter section 2 is as follows. To avoid significantly modifying the overall filter output impedance the section 2 output impedance must be sufficiently less than and It can be seen from Fig. 10.26 that, with respect to this is most difficult to accomplish when the peak frequencies of sections 1 and 2 coincide. It is most difficult to satisfy the design criterion when the peak frequency of sections 2 is lower than the peak frequency of section 1. Therefore, the best choice is to stagger-tune the filter sections, with the resonant frequency of section 1 being lower than the peak frequency of section 2. This implies that section 1 will produce more high-frequency attenuation than section 2. For this reason, we have chosen to achieve 45 dB of attenuation with section 1, and 35 dB of attenuation from section 2. The section 2 undamped resonant frequency should be chosen in the same manner used in Eq. (10.46) for section 1. We have chosen to select for section 2; this again means that the damping network will degrade the high frequency attenuation by a factor of or 9.5 dB. Hence, the section 2 undamped resonant frequency should be chosen to yield 35 dB + 9.5 of attenuation at 250 kHz. Since section 2 exhibits a two-pole ( 40 dB/decade) roll-off at high frequencies, should be chosen as follows:

The output impedance of section 2 will peak at the frequency 27.2 kHz, as given by Eq. (10.36). Hence, the peak frequencies of sections 1 and 2 differ by almost a factor of 2.

402

Input Filter Design

Figure 10.26 shows that, at 27.2 kHz, has a magnitude of roughly and that is approximately Hence, let us design section 2 to have a peak output impedance of Solution of Eq. (10.37) for the required section 2 characteristic impedance leads to

The section 2 element values are therefore

A Bode plot of the resulting is overlaid on Fig. 10.26. It can be seen that is less than, but very close to, between the peak frequencies of 15 kHz and 27 kHz. The impedance inequalities (10.45) are satisfied somewhat better below 15 kHz, and are satisfied very well at high frequency. The resulting filter output impedance is plotted in Fig. 10.27, for section 1 alone and for the complete cascaded two-section filter. It can be seen that the peak output impedance is approxi-

10.5

Summary of Key Points

403

mately or roughly The impedance design criteria (10.13) are also shown, and it can be seen that the filter meets these design criteria. Note the absence of resonances in The effect of stage 2 on is very small above 40 kHz [where inequalities (10.45) are very well satisfied], and has moderate-to-small effect at lower frequencies. It is interesting that, above approximately 12 kHz, the addition of stage 2 actually decreases The reason for this can be seen from Fig. C.8 of Appendix C: when the phase difference between and is not too large then the term decreases the magnitude of the resulting As can be seen from the phase plot of Fig. 10.26, this is indeed what happens. So allowing to be similar in magnitude to above 12 kHz was an acceptable design choice. The resulting filter transfer function is illustrated in Fig. 10.28. It can be seen that it does indeed attain the goal of 80 dB attenuation at 250 kHz. Figure 10.29 compares the single stage design of Section 10.4.1 to the two-stage design of this section. Both designs attain 80 dB attenuation at 250 kHz, and both designs meet the impedance design criteria of Eq. (10.13). However, the single-stage approach requires much larger filter elements.

10.5
1. 2. 3.

SUMMARY OF KEY POINTS

Switching converters usually require input filters, to reduce conducted electromagnetic interference and possibly also to meet requirements concerning conducted susceptibility. Addition of an input filter to a converter alters the control-to-output and other transfer functions of the converter. Design of the converter control system must account for the effects of the input filter. If the input filter is not damped, then it typically introduces complex poles and RHP zeroes into the converter control-to-output transfer function, at the resonant frequencies of the input filter. If these resonant frequencies are lower than the crossover frequency of the controller loop gain, then the phase margin will become negative and the regulator will be unstable.

404

Input Filter Design

References

405

4.

The input filter can be designed so that it does not significantly change the converter control-to-output and other transfer functions. Impedance inequalities (10.13) give simple design criteria that guarantee this. To meet these design criteria, the resonances of the input filter must be sufficiently damped. Optimization of the damping networks of single-section filters can yield significant savings in filter element size. Equations for optimizing three different filter sections are listed. Substantial savings in filter element size can be realized via cascading filter sections. The design of noninteracting cascaded filter sections can be achieved by an approach similar to the original input filter design method. Impedance inequalities (10.45) give design criteria that guarantee that interactions are not substantial.

5. 6.

REFERENCES

[1]

M. NAVE, Power Line Filter Design for Switched Mode Power Supplies, New York: Van Nostrand Reinhold, 1991. Design Guide for Electromagnetic Interference (EMI) Reduction in Power Supplies, MIL-HDBK-241B, U.S. Department of Defense, April 1 1981. C. M ARSHAM , The Guide to the EMC Directive 89/336/EEC, New York: IEEE Press, 1992. P. D EGAUQUE and J. H AMELIN , Electromagnetic Compatibility, Oxford: Oxford University Press, 1993. R. REDL, Power Electronics and Electromagnetic Compatibility, IEEE Power Electronics Specialists Conference, 1996 Record, pp. 15-21. P. R. WILLCOCK, J. A. F ERREIRA , J. D. V A N W YK , An Experimental Approach to Investigate the Generation and Propagation of Conducted EMI in Converters, IEEE Power Electronics Specialists Conference, 1998 Record, pp. 1140-1146. L. ROSSETTO, S. BUSO and G. SPIAZZI, Conducted EMI Issues in a 600W Single-Phase Boost PFC Design, IEEE Transactions on Industry Applications, Vol. 36, No. 2, pp. 578-585, March/April 2000. F. DOS R EIS , J. S EBASTIAN and J. U CEDA , Determination of EMI Emissions in Power Factor Preregulators by Design, IEEE Power Electronics Specialists Conference, 1994 Record, pp. 1117-1126. R. D. MIDDLEBROOK, Input Filter Considerations in Design and Application of Switching Regulators, IEEE Industry Applications Society Annual Meeting, 1976 Record, pp. 366-382. R. D. MIDDLEBROOK , Design Techniques for Preventing Input Filter Oscillations in Switched-Mode Regulators, Proceedings of Powercon 5, pp. A3.1 A3.16, May 1978. T. PHELPS and W. TATE, Optimizing Passive Input Filter Design, Proceedings of Powercon 6, pp. G 1.1G1.10, May 1979. Y. J ANG and R. E RICKSON , Physical Origins of Input Filter Oscillations in Current Programmed Converters, IEEE Transactions on Power Electronics, Vol 7, No. 4, pp. 725-733, October 1992. S. E RICH and W. P OLIVKA , Input Filter Design for Current-Programmed Regulators, IEEE Applied Power Electronics Conference, 1990 Proceedings, pp. 781-791, March 1990.

[2]

[3] [4] [5]

[6]

[7]

[8]

[9]

[10]

[11]

[12]

[13]

406

Input Filter Design

[14]

N. SOKAL, System Oscillations Caused by Negative Input Resistance at the Power Input Port of a Switching Mode Regulator, Amplifier, Dc/Dc Converter, or Dc/Ac Inverter, IEEE Power Electronics Specialists Conference, 1973 Record, pp. 138-140. A. K ISLOVSKI , R. REDL, and N. S OKAL , Dynamic Analysis of Switching-Mode Dc/Dc Converters, New York: Van Nostrand Reinhold, Chapter 10, 1991. S. K ELKAR and F. LEE, A Novel Input Filter Compensation Scheme for Switching Regulators, IEEE Power Electronics Specialists Conference, 1982 Record, pp. 260-271. R. E RICKSON , Optimal Single-Resistor Damping of Input Filters, IEEE Applied Power Electronics Conference, 1999 Proceedings, pp. 1073-1097, March 1999. M. F LOREZ -L IZARRAGA and A. F.W ITULSKI , InputFilter Design for Multiple-Module DC Power Systems, IEEE Transactions on Power Electronics, Vol 11, No. 3, pp. 472-479, May 1996. and F. LEE, Input Filter Design for Power Factor Correction Circuits, IEEE Transactions on Power Electronics, Vol 11, No.1, pp. 199-205, January 1996. F. Y UAN , D. Y. C HEN , Y. WU and Y. C HEN , A Procedure for Designing EMI Filters for Ac Line Applications, IEEE Transactions on Power Electronics, Vol 11, No. 1, pp. 170-181, January 1996. G. SPIAZZI and J. P OMILIO , Interaction Between EMI Filter and Power Factor Preregulators with Average Current Control: Analysis and Design Considerations, IEEE Transactions on Industrial Electronics, Vol. 46, No. 3, pp. 577-584, June 1999.

[15]

[16]

[17]

[18]

[19]

[20]

[21]

PROBLEMS

10.1

It is required to design an input filter for the flyback converter of Fig. 10.30. The maximum allowed rms. Calculate the required attenuation of the filter at amplitude of switching harmonics of is the switching frequency.

10.2

In the boost converter of Fig. 10.31, the input filter is designed so that the maximum amplitude of switching harmonics of is not greater than rms. Find the required attenuation of the filter at the switching frequency. Derive the expressions for and in Table 10.1.

10.3

Problems

407

10.4

The input filter for the flyback converter of Fig. 10.30 is designed using a single is damped using a resistor in series with a very large blocking capacitor (a) (b)

section. The filter

Sketch a small-signal model of the flyback converter. Derive expressions for and using your model. Sketch the magnitude Bode plots of and and label all salient features. Design the input filter, i.e., select the values of and , so that: the filter attenuation at the switching frequency is at least 100 dB, and (ii) the magnitude of the filter output impedance satisfies the conditions for all frequencies. Use Spice simulations to verify that the filter designed in part (b) meets the specifications. Using Spice simulations, plot the converter control-to-output magnitude and phase responses without the input filter, and with the filter designed in part (b). Comment on the changes introduced by the filter. section.

(c) (d)

10.5

It is required to design the input filter for the boost converter of Fig. 10.31 using a single The filter is damped using a resistor in series with a very large blocking capacitor (a) (b) Sketch the magnitude Bode plots of features.

and

for the boost converter, and label all salient

and , so that: the filter attenuation at Design the input filter, i.e., select the values of the switching frequency is at least 80 dB, and (ii) the magnitude of the filter output impedance satisfies the conditions for all frequencies.
Use Spice simulations to verify that the filter designed in part (b) meets the specifications. Using Spice simulations, plot the converter control-to-output magnitude and phase responses without the input filter, and with the filter designed in part (b). Comment on the changes in the control-to-output responses introduced by the filter.

(c) (d)

10.6 10.7 10.6 10.9

Repeat the filter design of Problem 10.4 using the optimum filter damping approach described in Section and 10.4.1. Find the values of Repeat the filter design of Problem 10.5 using the optimum filter damping approach of Section 10.4.1. and Find the values of Repeat the filter design of Problem 10.4 using the optimum and in Section 10.4.2. Find the values of Repeat the filter design of Problem 10.5 using the optimum and in Section 10.4.2. Find the values of parallel damping approach described parallel damping approach described

408

Input Filter Design

10.10

It is required to design the input filter for the flyback converter of Fig. 10.30 using two filter sections. Each filter section is damped using a resistor in series with a blocking capacitor. (a) Design the input filter, i.e., select values of all circuit parameters, so that (i ) the filter attenuation at the switching frequency is at least 100 dB, and (ii) the magnitude of the filter output impedance satisfies the conditions for all frequencies. Use Spice simulations to verify that the filter designed in part (a) meets the specifications. Using Spice simulations, plot the converter control-to-output magnitude and phase responses without the input filter, and with the filter designed in part (b). Comment on the changes introduced by the filter.

(b) (c)