IECON2015-Yokohama

November 9-12, 2015

Small-Signal Analysis of PWM Boost Converter in

CCM with Complex Impedance Load
Dalvir K. Saini, Agasthya Ayachit, Tadashi Suetsugu
and Marian K. Kazimierczuk Department of Electronics Engineering
Department of Electrical Engineering and Computer Science
Wright State University Fukuoka University
Dayton, Ohio 45435, USA Fukuoka, 814-0180, Japan
Email: {saini.11, ayachit.2, marian.kazimierczuk}@wright.edu Email: suetsugu@fukuoka-u.ac.jp

Abstract—This paper presents a small-signal analysis of the

pulse-width modulated (PWM) boost dc-dc converters operating > Ϭ
in continuous-conduction mode (CCM) and subject to complex
impedance load. Using the existing knowledge on the small-
Z>
signal model of the boost converter obtained through the circuit
s/ н
averaging approach, this paper presents an extended analysis
of the small-signal transfer functions, under the inuence of
^  ǀK
resistive and inductive load impedances. The expressions for the >> ͺ
control-to-output, input-to-output voltage transfer functions, and ǀ'^
the output impedance of the boost converter operating in CCM
are derived. The characteristics of these transfer functions are
analyzed and the effect of the added inductance on the location
of poles and zeros is discussed. It is shown that the inherent Fig. 1. Circuit of the boost converter with inductive load.
right-half plane (RHP) zero shifts to the left-half of the s−plane,
when the load inductance is increased beyond a specic value.
The theoretical results are validated by simulations of a boost constant power to a cascade-connected PWM converter or dif-
converter with suitable design specications.
ferent types of nonlinear loads. Various methods to determine
the dynamic properties of the converters are also presented.
I. I NTRODUCTION While these contributions provide an in-depth understanding
Pulse-width modulated dc-dc converters have found their of the complete set-up comprising of the main power supply
importance as highly efcient power supplies for various types and the cascaded power converter, the present paper aims at
of circuits (or output stages) such as series-connected power modeling only the main power stage and treat the series-
converters, radio-frequency power ampliers, dc motors, etc. connected impedance as a simple inductive or capacitive load.
In such systems, the load presented to the primary power The objectives of this paper are as follows:
supply is no longer purely resistive. This problem is evident in • To adopt the load inductance model to an already existing
amplitude-modulated communication systems, where the buck small-signal model of the boost converter with resistive
or boost power converter is connected to a Class-E RF power load.
amplier. In such situations, the effective load resistance to the • To develop the different transfer functions namely
power converter is a series combination of the load resistance control-to-output Tp , the input-to-output voltage Mv , and
as seen by the converter and the choke inductance at the input the expression for the output impedance of the boost
of the power amplier. Similar problems exist in the case of converter in CCM.
main power supplies loaded with successive power converter • To analyze the effect of different values of the load
stages such as in point-of-load applications or multiple-output inductance on the location of poles and zeros of the
converter stages. Thus, the dynamics of the power supply, derived transfer functions.
when subject to change in the load, control voltage, or duty
cycle must be carefully investigated.
II. S MALL -S IGNAL M ODEL OF THE B OOST C ONVERTER
The analyses presented in [1] - [8] addresses several chal- WITH C OMPLEX I MPEDANCE L OADS
lenges in designing and controlling the main voltage supply
delivering power to successively loaded converters. In [1], the Fig. 1 shows the circuit of the boost converter under
authors model the power stage in terms of a two-port network, consideration. The load is presented with an impedance, where
which accommodates the effect of dynamically changing out- RL is the dc load resistance, while LL is the load inductance.
put load impedance. In [2] - [8], the authors treat the main Using the techniques presented in [9] - [12], the small-signal
power supply (also called as the “line conditioner”) to deliver model of the boost converter is as shown in Fig. 2. The model

978-1-4799-1762-4/15/$31.00 ©2015 IEEE 003597

 ǀŽ sKĚ ϭ
> ƌ ǀƚ
н
> ƌ
ŝů Z>
н ƌ ŝƚ
Ě ǀŽ ŝŽ ŝů Z>
ǀŝ ƌ
ͺ />Ě ŝů  >> н
ͺ ŝů ǀƚ
 >> ͺ
Fig. 2. Complete small-signal equivalent model of the boost converter in
CCM.
Ϯ
ϭ
ǀŽ s KĚ Fig. 5. Small-signal model of the boost converter needed to determine the
> ƌ expression for the output impedance.

н
ŝů Z>
ƌ A. Control-to-Output Transfer Function Tp
Ě ǀŽ
/ >Ě ŝů Fig. 3 can be used to derive the control-to-output voltage
 >> transfer function of the boost converter with impedance load.
ͺ
The model is obtained by setting vi and io to zero. The
impedances in the model can be lumped and expressed as
Ϯ follows. Let
Z1 = r + sL, (1)
Fig. 3. Small-signal model of the boost converter needed to determine the
1
control-to-output transfer function. Z2 = (rC + )||(RL + sLL ). (2)
sC
Using the analysis presented in [10], [12], the small-signal
ϭ control-to-output voltage transfer function in terms of the
ǀŽ impedances can be determined as
> ƌ 
vo (s) 
н Tp (s) =
ŝů Z> d(s) vi =0, io =0
ƌ (3)
ǀŝ ǀŽ VO (1 − D)2 RL Z2 − Z1 Z2
= ,
 1 − D (1 − D)2 RL Z2 + RL Z1
ŝů >>
ͺ Substituting (1) and (2) into (3), we obtain the higher-order
control-to-output transfer as

Ϯ (s + ωzn ){s − ωzp }

Tp (s) = Tpx , (4)
s3 + αs2 + βs + γ
Fig. 4. Small-signal model of the boost converter needed to determine the where the high-frequency gain is
input-to-output voltage transfer function.
VO rC [L − (1 − D)2 LL ]
Tpx = Tp (∞) = − , (5)
1−D L(RL + rC )
of the boost converter is valid for operation in continuous- the zero at the left-half plane is
conduction mode (CCM). The model is obtained using the 1
circuit averaging technique, where the switch is replaced by a ωzn = , (6)
CrC
voltage-dependent current source and the diode is replaced by
a voltage-dependent voltage source [9], [10]. From the gure, and is completely independent of the value of LL . The LL
r forms the equivalent averaged resistance of the switch, diode, dependent right-half plane (RHP) zero of the transfer function
and inductor considered in series in the inductor branch, L is is
RL (1 − D)2 − r
the boost inductor, C is the lter capacitor, whose equivalent ωzp = , (7)
L − (1 − D)2 LL
series resistance is rC . In the following sections, the effect
of the inductance LL on dynamic performance of the boost the coefcients of the transfer function denominator are
converter will be analyzed with the aid of suitable transfer (1 − D)2 LL CrC + rLL C + LCRL + LCrC
functions. α= , (8)
LCLL

003598
 (1 − D)2 RL CrC + (1 − D)2 LL + CrrC + L while ωzn and ω1 are as given in (6) and (12), respectively,
β= , (9)
LCLL and the coefcients α, β, and γ are as given in (8), (9), (10),
(1 − D)2 RL respectively. Similar to the approximated control-to-output
γ= . (10) transfer function Tp given in (11), the denominator of Mv
LCLL
can be factorized to obtain
The derivation of the exact solution for the third-order trans-
fer function is cumbersome. However, a valid approximation (s + ωzn )(s + ω1 )
Mv (s) ≈ Mvx , (18)
can be made and a representative transfer function can be (s + ω1 )(s2 + 2ξω0 s + ω02 )
obtained, thanks to the extensive polynomial solving tool on
MATLAB. The resulting control-to-output transfer function is yielding
s + ωzn
(s + ωzn )(s − ωzp ) Mv (s) = Mvx . (19)
Tp (s) ≈ Tpx (11) s2 + 2ξω0 s + ω02
(s + ω1 )(s2 + 2ξω0 s + ω02 )
The introduction of LL results in an additional zero in the
where
RL input-to-output voltage transfer function. The location of the
ω1 = , (12) zero is always in the left-half of the s-plane.
LL

(1 − D)2 RL + r C. Output Impedance Zo
ω0 ≈ , (13)
LC(RL + rC )
Fig. 5 shows the small-signal model needed to determine
and the output impedance of the boost converter in CCM. Based
C[r(RL + rC ) + (1 − D)2 RL rC ] + L
ξ≈  . (14) on the discussion provided in [10], the output impedance in
2 LC(RL + rC )[r + (1 − D)2 RL ] the presence of load inductance is expressed as
From (5)-(10), it can be observed that the transfer function 
vt (s) 
Tp has a strong dependence on the value of the Inductance LL . Zo =
For smaller values of the load inductance, the zero ωzp exists it (s) d=0, io =0, vi =0
(20)
on the right-half of the s-plane, while for higher values of the Z 1 Z2
= .
inductance LL , the zero shifts to the left half-plane altering Z1 + Z2 (1 − D)2
the dynamics of the boost converter. The inductance LL does
not affect the high-frequency gain Tpx . The addition of the Substituting for Z1 and Z2 given in (1) and (2) into (20), we
load inductance LL introduces an extra pole in the control- get
to-output transfer function. As LL decreases, the location of (s + ωzn )(s + ω1 )(s + ω2 )
Zo (s) = Zox , (21)
the pole moves towards −∞. As LL increases, the ringing s3 + αs2 + βs + γ
in its step response also increases. Detailed analysis of the
control-to-output transfer function will be discussed in Section where the high-frequency gain is
III along with suitable design example. rC RL
Zox = Zo (∞) = ≈ rC , (22)
B. Input-to-Output Voltage Transfer Function Mv rC + RL
The small-signal model of the boost converter needed to the zero due to the combination of r and the L is
determine the input-to-output voltage transfer function Mv is
r
shown in Fig. 4. Using the analysis presented in [10], [12], ω2 = , (23)
the expression for the input-to-output voltage transfer function L
in terms of the impedances is given by and ωzn , ω1 , α, β, and γ are as given in (6), (12), (8), (9), and

vo (s)  (10), respectively. The denominator of (21) can be factorized
Mv (s) =
vi (s) d=0, io =0 to yield
(15)
Z2 (1 − D) (s + ωzn )(s + ω1 )(s + ω2 )
= , Zo (s) = Zox , (24)
Z1 + Z2 (1 − D)2 (s + ω1 )(s2 + 2ξω0 s + ω02 )
where the impedances Z1 and Z2 are as given in (1) and (2),
resulting in
respectively. Substituting for the impedances, we get
(s + ωz )(s + ω1 ) (s + ωzn )(s + ω2 )
Mv (s) = Mvx , (16) Zo (s) = Zox . (25)
s3 + αs2 + βs + γ s2 + 2ξω0 s + ω02
where the high-frequency gain of the Mv transfer function is The load impedance LL adds an additional pole-zero pair
RL rC (1 − D) to the expression for the output impedance. The location of
Mvx = Mv (∞) = , (17) the pole-zero pair is always in the left-half of the s-plane.
L(RL + rC )

003599
 III. R ESULTS AND D ISCUSSION
A boost converter is designed for the following speci- 50

cations: input dc voltage VI = 12 V, switching frequency

fs = 100 kHz, minimum output power POmax = 10 W,
and the dc output voltage is VO = 20 V. Using the design

| Tp| (dB)
0
equations presented in [10], the values of the boost inductor
and capacitors for a nominal duty ratio of D = 0.46 are
found to be: L = 156 μH, and C = 6.8 μF. The equivalent
average resistance (EAR) considered in the inductor branch 50
0
is r = 0.24 Ω. The equivalent series resistance of the lter 45
capacitor is rC = 0.111 Ω. 90

φ Tp (deg)
The following discussion was considered to assign suitable 135 LL = 53 μH
values for the load inductance. From the control-to-output 180 LL = 0.53 mH
transfer function, consider the expression for the right-half 225 LL = 5.3 mH
plane (RHP) zero as given in (7). For a xed value of the 270
3 4 5
10 10 10
load resistance RL and the duty cycle D, the location of the f (Hz)
RHP zero is
Fig. 6. Theoretically obtained Bode plots of the control-to-output transfer
⎧ function Tp at the three different values of the load inductance LL .
⎨ right-half plane for LL < L/(1 − D)2
ωzp = zp = +∞ for LL = L/(1 − D)2
⎩
left-half plane for LL > L/(1 − D)2 . oscillations. The validity of the presented theoretical results
(26) is tested by simulating the given boost converter for the three
Using the conditions given in (26) and the given values of inductance on the Saber circuit simulator. Fig. 9 shows the
L and D, the load inductance resulting in zp = ∞ is LL = Bode plot of the control-to-output voltage transfer function
0.53 mH. To satisfy the other conditions, the assumed values obtained through simulations. The theoretical and simulation
of LL are 53 μH and 5.3 mH, respectively. results are in agreement with each other for most of the useful
frequency range.
A. Analysis of Control-to-Output Transfer Function Tp
B. Analysis of Audio Susceptibility Mv
The frequency-domain characteristics of the control-to-
output transfer function will be analyzed in this section. Fig. Fig. 10 shows the Bode magnitude and phase plots of the
6 shows the Bode magnitude and phase plot of the Tp transfer input-to-output voltage transfer function. Similar to Tp , the
function for the three different values of the load inductance. effect of the load inductance on the Mv transfer function is
For all the three inductances, the values of the dc and low- insignicant for LL ≤ L/(1 − D)2 = 0.53 mH. For higher
frequency gain is the same and is equal to 31 dB. Similarly, values of LL , i.e., LL > 0.53 mH, it was observed that the
the corner frequency ω0 at which the peak occurs also remains damping coefcient of the second-order transfer function in
unchanged and has a value ω0 = 2.6 kHz. (19) is approximately ξ = 0.068, causing a peak in the Bode
Fig. 7 shows the representative root-locus plot of the magnitude plot and an oscillatory nature in the step response
control-to-output transfer function for the twe extreme values plot. Fig. 11 shows the Bode plot of the input-to-output voltage
of the load inductance. Using (26), for LL = 53 μH, the transfer function obtained through simulations. It can be seen
sign of the zero ωzp is positive and lies on the RHP. For that the theoretical and simulation results are in agreement
LL = 5.3 mH, the sign changes causing the zero to relocate with each other for the useful frequency range.
to the left-half of the s−plane. Similarly, the location of the C. Analysis of Output Impedance Zo
pole at ω1 also shifts towards the origin for higher values of Fig. 12 shows the Bode magnitude and phase plot of the
LL , causing the converter to exhibit an oscillatory response. output impedance for three different values of the load induc-
Using MATLAB’s control system toolbox, the damping tance. Similar to the analysis on the input-to-output voltage
coefcient of the Tp transfer function was found to be ξ = transfer function, the characteristics of Zo remains unaltered
0.1632 for LL ≤ 0.53 mH and ξ = 0.068 for LL = 5.32 mH. for LL ≤ 0.53 mH. However, for LL > 0.53 mH, the peak
The effect of the low value of the damping coefcient at occurs due to smaller values of the damping coefcient. The
LL = 5.32 mH can be observed through the overshoot in result is also veried using the simulated Bode plot as shown
the Bode magnitude plot and also the oscillatory nature of the in Fig. 13. Finally, Fig. 14 shows the plot of the output voltage
output voltage in the step response plot shown in Fig. 8. The for step changes in the load current.
effect of the left-half plane zero namely ωzn2 especially for
LL ≥ 0.53 mH can be observed at high frequencies, where IV. C ONCLUSION
the phase of the Bode plot is close to −180◦ at the cross-over This paper has presented a small-signal analysis of
frequency, again causing the output voltage to exhibit damped the pulse-width modulated boost dc-dc converter with an

003600
 ũʘ Mag_Tp (dB/V) : f(Hz)

Tp@LL=53 uH

ʘ 

Mag_Tp (dB/V)
40.0

20.0 Tp@LL=5.3 mH
ʘ
 ʘ
 0.0

ʍ
ʘ Ph_Tp (deg) : f(Hz)

0.0
Tp@LL=53 uH

Ph_Tp (deg)
ʘ −100.0
Tp@LL=5.3 mH
;ĂͿ −200.0

ũʘ −300.0

1.0k 10.0k 100.0k

ʘ  f(Hz)

ʘ
 ʘ
 Fig. 9. Simulated magnitude and phase Bode plots of the control-to-output
voltage transfer function Tp at the three different values of the load inductance
ʍ LL .
ʘ

ʘ
;ďͿ
20
Fig. 7. Pole-zero plot for the control-to-output voltage transfer function Tp
| M | (dB)

for two different values of the load inductance, where ωp1 , ωp2 represent 0
the poles of the second-order polynomial in (11). (a) Location of poles and L = 53 μH
v

20
zeros, when LL < L/(1 − D)2 . (b) Location of poles and zeros, when L
L = 0.53 mH
LL > L/(1 − D)2 . 40
L
L = 5.3 mH
L
60
0
26
LL = 53 μH 45
(deg)

25 LL = 0.53 mH
90
v

LL = 5.3 mH
M
φ

24 135

180
23 2 3 4 5
vO (V)

10 10 10 10
f (Hz)
22
Fig. 10. Theoretically obtained Bode plots of the input-to-output voltage
21 transfer function Mv at the three different values of the load inductance LL .

20

19
characteristics. A few critical and veriable features have been
0 0.5 1 1.5 2 2.5 3 observed. The addition of the load inductance introduces an
t (ms) extra pole in the Tp transfer function, an extra zero in the Mv
Fig. 8. Theoretical response of the output voltage to a step change in the
transfer function, and an extra pole-zero pair in the output
duty cycle (ΔD = 0.1) at three different values of the load inductance LL . impedance Zo as shown in (11), (19), and (25). As the load
inductance increases: (a) the location of the inherent right-half-
plane (RHP) zero in the control-to-output transfer function
impedance load. The dynamics of the converter in the presence Tp moves closer to the origin and enters the left-half of the
of a series-connected resistance and inductance as the load s−plane (see Fig. 7) and (b) the amplitude and the settling time
have been presented. The small-signal transfer functions such of the ringing in the output voltage increase and are shown in
as the control-to-output transfer function Tp , the input-to- the step response plots (see Fig. 8 and Fig. 14). Future work
output voltage transfer function Mv , and the output impedance constitutes the analysis in the presence of capacitive load and
Zo have been derived. Analysis of these expressions have the analysis of the effect of reactive load on the closed-loop
been made using their frequency-domain and time-domain small-signal transfer functions.

003601
 Mag_Mv (dB) : f(Hz) Mag_Zo (dBohm) : f(Hz)
20.0
40.0
Mv@LL=53 uH Zo@LL=53 uH

Mag_Zo (dBohm)
30.0
Mag_Mv (dB)

0.0

Mv@LL=5.3 mH 20.0 Zo@LL=5.3 mH

−20.0
10.0
−40.0
0.0
−60.0 −10.0
Ph_Mv (deg) : f(Hz) Ph_Zo (deg) : f(Hz)

0.0
50.0
Mv@LL=53 uH Zo@LL=53 uH
Ph_Mv (deg)

Ph_Zo (deg)
−50.0

Mv@LL=5.3 mH Zo@LL=5.3 mH
0.0

−100.0

−50.0
−150.0

−100.0
−200.0

1.0k 10.0k 100.0k 1.0k 10.0k 100.0k

f(Hz) f(Hz)

Fig. 11. Simulated magnitude and phase Bode plots of the input-to-output Fig. 13. Simulated Bode plot of the output impedance Zo at the two different
voltage transfer function Mv at the two different values of the load inductance values of the load inductance LL .
LL .

30
LL = 53 μH
40 28
LL = 0.53 mH
30 LL = 5.3 mH
26
20
| Zo| (dB)

10 24
vO (V)

0
22
10

20 20
90
LL = 53 μH
18
45
L = 0.53 mH
(deg)

0 L = 5.3 mH 16
L
o
Z
φ

45 14
0 0.5 1 1.5 2 2.5 3
90 t (ms)
2 3 4 5
10 10 10 10
f (Hz) Fig. 14. Theoretical response of the output voltage to a step change in the
load current (ΔIO = 0.1 A) at three different values of the load inductance.
Fig. 12. Theoretically obtained Bode plot of the output impedance Zo at the
three different values of the load inductance LL .
systems,” IEEE Trans. Aerosp. Electron. Syst., vol. 36, no. 3, pp. 965-
973, July 2000.
R EFERENCES [7] U. Borup, F. Blaabjerg, and P. N. Enjeti, ”Sharing of nonlinear load in
parallel-connected three-phase converters,” IEEE Trans. Ind. Appl., vol.
[1] M. Hankaniemi, T. Suntio, M. Sippola, and E. Oyj, ”Load-impedance 37, no. 6, pp. 1817-1823, Dec. 2001.
based interactions in regulated converters,” in Proc. in 27th IEEE Inter- [8] J. M. Guerrero, L. G De Vicuna, J. Matas, M. Castilla, and J. Miret,
national Telecommunications Conf., INTELEC’05, Berlin, Germany, Sept. ”Output impedance design of parallel-connected UPS inverters with
2005, pp. 569-573. wireless load-sharing control,” IEEE Trans. Ind. Electron., vol. 52, no. 4,
[2] B. Choi, J. Kim, B. H. Cho, S. Choi, and C. M. Wildrick, ”Designing con- pp. 1126-1135, Aug. 2005.
trol loop for DC-to-DC converters loaded with unknown AC dynamics,” [9] D. Czarkowski and M. K. Kazimierczuk, “Energy-conservation approach
IEEE Trans. Ind. Electron., vol. 49, no. 4, pp. 925-932, Aug. 2002. to modeling PWM DC-DC converters,” IEEE Trans. Aerosp. Electron.
[3] L. Peng and B. Lehman, “Performance prediction of DC-DC converters Syst., vol. 29, no. 3, pp. 1059-1063, Jul. 1993.
with impedances as loads,” IEEE Trans. Power Electron., vol. 19, no. 1, [10] M. K. Kazimierczuk, Pulse-width Modulated dc-dc Power Converters,
pp. 201-209, Jan. 2004. 2nd Ed., John Wiley & Sons, Chichester, United Kingdom, 2015.
[4] C. Byungcho, B. H. Cho, and H. Sung-Soo, “Dynamics and control of [11] B. Bryant and M. K. Kazimierczuk, “Voltage-loop power-stage transfer
DC-to-DC converters driving other converters downstream,” IEEE Trans. functions with MOSFET delay for boost PWM converter operating in
Circ. Syst-I: Fundam. Theory. Appl., vol. 46, no. 10, pp. 1240-1248, Oct. CCM,” IEEE Trans. Ind. Electron., vol. 54, no. 1, pp. 347-353, Feb.
1999. 2007.
[5] F. Xiaogang, L. Jinjun, and F. .C. Lee, ”Impedance specications for [12] B. Bryant and M. K. Kazimierczuk, “Voltage loop of boost PWM DCDC
stable DC distributed power systems,” IEEE Trans. Power Electron., vol. converters with peak current-mode control,” IEEE Trans. Circ. Systems-I:
17, no. 2, pp. 157-162, Mar. 2002. Regular Papers, vol. 53, no. 1, pp. 99-107, Jan. 2006.
[6] S. D. Sudhoff, S. F. Glover, P. T. Lamm, D. H. Schmucker, and
D. E. Delisle,”Admittance space stability analysis of power electronic

003602