IEEE TRANSACTIONS ON INDUSTRIAL ELECTRONICS, VOL. 54, NO.

1, FEBRUARY 2007 347

Voltage-Loop Power-Stage Transfer Functions With

MOSFET Delay for Boost PWM Converter
Operating in CCM
Brad Bryant, Member, IEEE, and Marian K. Kazimierczuk, Fellow, IEEE

Abstract—Open-loop transfer functions can be used to create

closed-loop models of pulsewidth-modulated (PWM) converters.
The closed-loop small-signal model can be used to design a con-
troller for the switching converter with well-known linear control
theory. The dynamics of the power stage for boost PWM dc–dc
converter operating in continuous-conduction mode (CCM) are
studied. The transfer functions from output current to output
voltage, from duty cycle to output voltage including MOSFET
delay, and from input voltage to output voltage are derived.
The derivations are performed using an averaged linear circuit
small-signal model of the boost converter for CCM. Experimental
Bode plots and step responses were used to test the accuracy of
the derived transfer functions. The theoretical and experimental Fig. 1. Open-loop boost PWM converter.
responses were in excellent agreement, confirming the validity of
the derived transfer functions.
impedance in both current-mode and voltage-mode control of
Index Terms—Boost converter, delay, pulsewidth modulated PWM converters. The relationship between duty cycle and
(PWM), PWM power converters, step responses, switch-mode output voltage is essential in voltage-mode control and is
power supply, transfer functions.
important for describing the voltage loop for current-mode
control of PWM converters. The transfer function from input
I. I NTRODUCTION
voltage to output voltage, or open-loop audio susceptibility,
is important to determine the closed-loop audio susceptibil-
M ODELING the dynamic behavior of nonlinear
pulsewidth-modulated (PWM) dc–dc converters has
been studied in previous literature [1]–[11]. State-space
ity in both current-mode and voltage-mode control of PWM
converters. The transfer functions from duty cycle to output
averaging is one method presented in [1]–[5]. This method voltage, output current to output voltage, and input voltage to
can be tedious due to matrix algebra manipulations. Another output voltage can be determined using an averaged linear time-
method is based on equivalent circuit modeling [1]–[11]. invariant circuit model of the nonlinear converter [8], [9].
The controlled source model in [1] accounts for inductor and The objectives of this paper are to derive the small-signal
capacitor losses but does not account for the transistor and output current to output-voltage transfer function Zo , the small-
diode conduction losses. The equivalent circuit model in [6] for signal duty cycle to output voltage transfer function Tp , and the
continuous-conduction mode (CCM) uses a different model small-signal input voltage to output voltage transfer function
topology than presented in [8], [9] and no model verification Mv , and to give experimental verification using as an example
was given. The controlled source model in [8] and [9] accounts a boost PWM dc–dc converter operating in CCM.
for the transistor and diode conduction losses but does not
account for MOSFET delay. The advantages of equivalent II. M ODELING THE PWM B OOST C ONVERTER IN CCM
circuit modeling are simplicity and clearer insight into con-
verter behavior. A circuit averaging technique can be used to model PWM
Open-loop transfer functions can be used to create closed- converters. In PWM converters, transistors and diodes are the
loop models of PWM converters using linear system theory. prevalent switching devices. Fig. 1 shows an open-loop boost
Linear models allow the use of classic linear control tech- PWM converter with small-signal disturbance input voltage vi ,
niques for controller design purposes [3]. The transfer function disturbance output current io , and duty cycle d as a control
from output current to output voltage, or open-loop output signal. The boost circuit is used as a case study in this paper.
impedance, is important to determine the closed-loop output For large-signals, the transistor can be modeled as a current-
dependent current source as [8], [9]
Manuscript received December 1, 2004; revised March 17, 2005. Abstract
published on the Internet September 15, 2006. iS = dT iL (1)
The authors are with the Department of Electrical Engineering, Wright
State University, Dayton, OH 45435 USA (e-mail: [email protected];
[email protected]). where the total duty cycle dT = D + d, the switch current
Digital Object Identifier 10.1109/TIE.2006.885136 iS = IS + is , and inductor current iL = IL + il . D, IS , and

0278-0046/$25.00 © 2007 IEEE

348 IEEE TRANSACTIONS ON INDUSTRIAL ELECTRONICS, VOL. 54, NO. 1, FEBRUARY 2007

Fig. 2. Large-signal linear model of boost PWM converter operating in CCM.

Fig. 3. Small-signal model of boost PWM converter operating in CCM.

IL are the dc components of duty cycle, switch current, and

Fig. 4. Small-signal model to determine Zo for boost converter operating
inductor current, respectively. The small-signal ac components in CCM.
of duty cycle, switch current, and inductor current are d, is , and
il , respectively. For large signals, the diode can be modeled as
a voltage-dependent voltage source in CCM operation as the small-signal ac linear circuit model of a boost converter
operating in CCM. The small-signal model can be used to find
vD = d T vO (2) the transient behavior and transfer functions from input d to
outputs vo and il , and from disturbance inputs vi and io to
where vD = VD + vd , and vO = VO + vo . VO and VD are the outputs vo and il . The small-signal model in Fig. 3 is valid from
dc components of output and diode voltages, respectively. The dc to half of the converter switching frequency fs , which is the
small-signal ac components of output and diode voltages are vo Nyquist frequency.
and vd , respectively. The resulting transistor and diode models
become
III. O UTPUT I MPEDANCE
iS = (D + d)(IL + il ) = DIL + Dil + IL d + il d (3)
The open-loop output impedance, or output current to output-
and voltage transfer function Zo , can be found by setting control
input d and disturbance input vi of Fig. 3 to zero. A test voltage
vD = (D + d)(VO + vo ) = DVO + Dvo + VO d + vo d. (4) vt is then applied to the output, resulting in the circuit of Fig. 4.
Z1 and Z2 are
The double small-signal terms il d and vo d can be neglected
when the following small-signality conditions are met: il
IL ,
Z1 = sL + r (6)
vo
VO , and d
D. This enables linearization of the non-
linear model and derivation of small-signal transfer functions.
Fig. 1 shows a boost PWM converter. In Fig. 1, rL is the and
inductor equivalent series resistance (ESR), rc is the capacitor

1
ESR, and RL is the converter load resistance. Fig. 2 shows Z2 = + rC RL . (7)
a linear large-signal boost converter model operating in CCM sC
with the transistor and diode replaced with their respective
models. In Fig. 2, r was derived using the principle of energy Using KCL
conservation [8], [9] as
il − Dil − i = 0
r = rL + DrDS + (1 − D)RF (5)
and
where rDS is the transistor ON resistance, and RF is the diode
forward resistance. The large-signal linear model of Fig. 2 vt
i = iZ2 − it = − it
can be separated into dc and ac linear models. Fig. 3 shows Z2
BRYANT AND KAZIMIERCZUK: VOLTAGE-LOOP POWER-STAGE TRANSFER FUNCTIONS WITH MOSFET DELAY 349

Fig. 7. Zo Bode plots from Hewlett-Packard 4194A Impedance/Gain-Phase

Analyzer.
Fig. 5. Zo magnitude plot.

Fig. 8. Step response of vO due to ∆io = 0.1 A.

Fig. 6. Zo phase response. Substituting for Z1 and Z2 gives

LCrC RL
which gives Zo (s) =
LC(rC +RL )
  
vt it s+ Lr s+ Cr1C
il = − . (8)
Z2 (1 − D) (1 − D) ×  2
 2 R +r
s2 + C[r(RL +rLC(R
C )+RL rC (1−D) ]+L
L +rC )
s+ (1−D) L
LC(RL +rC )

From KVL, we get (s+ωrl )(s+ωzn )

= Zox (10)
s2 +2ζωn s+ωn2
−il Z1 + Dvt − vt = 0
where
yielding rC RL
Zox = = rC RL (11)
RL + rC
−vt (1 − D) r
il = . (9) ωrl = (12)
Z1 L
1
Using (8) and (9) results in the output impedance Zo as ωzn = (13)
CrC

vt vo Z1 Z2 (1 − D)2 RL + r
Zo = |v =d=0 = = . ωn = (14)
it i −io Z1 + Z2 (1 − D)2 LC(RL + rC )
350 IEEE TRANSACTIONS ON INDUSTRIAL ELECTRONICS, VOL. 54, NO. 1, FEBRUARY 2007

Fig. 9. Experimental step response of vO due to ∆io = 0.1 A. Upper trace

is ∆vo ; lower trace is ∆io . Horizontal scale: 2 ms/div. Vertical scale for upper
trace is 100 mV/div.; vertical scale for lower trace is 50 mA/div.

Fig. 11. Tp magnitude response.

Fig. 10. Small-signal model to determine Tp for boost converter operating

in CCM.

and

C r(RL + rC ) + RL rC (1 − D)2 + L
ζ= . (15)
2 LC(RL + rC ) [r + (1 − D)2 RL ]

The dc output resistance Zo (0) is

Fig. 12. Tp phase response. Solid line is for Td = 0, and dashed line is for
Td = 1 µs.
rRL
Zo (0) = . (16)
(1 − D)2 RL + r IV. D UTY C YCLE TO O UTPUT V OLTAGE
T RANSFER F UNCTION
For an example open-loop boost converter with a switch-
ing frequency fs = 100 kHz, VI = 10 V, RL = 50 Ω, D = The duty cycle to output voltage transfer function Tp can be
0.5, L = 254 µH, and C = 69 µF, the output voltage found by setting disturbance inputs vi and io of Fig. 3 to zero,
was VO = 18.47 V. The efficiency was 92.35%. The para- resulting in the circuit of Fig. 10. Tp (s) is derived as
sitics rL = 0.15 Ω and rC = 0.311 Ω were measured using
vo (s) (s − ωzp )(s + ωzn )
the Hewlett-Packard 4194A Spectrum Analyzer. The para- Tp (s) = |v =i =0 = Tpx 2 (17)
sitics rDS = ∆VDS /∆ID = 0.5 V/1 A = 0.5 Ω, and RF = d(s) i o s + 2ζωn s + ωn2
∆VD /∆ID = 0.245 V/0.9 A = 0.272 Ω were calculated from where
the measured static device characteristics. We calculate Zox =
0.3091 Ω, Zo (0) = 2.056 Ω, r = 0.536 Ω, frl = 335.9 Hz, −rC VO
Tpx = (18)
fzn = 7.417 kHz, fn = 611.95 Hz, and ζ = 0.3514. The exper- (RL + rC )(1 − D)
imental circuit switches were an IRF520 power MOSFET and
U860 power diode. The theoretical and experimental Bode plots and
of Zo for this example are shown in Figs. 5–7, respectively, and RL (1 − D)2 − r
were in excellent agreement. Figs. 8 and 9 show theoretical and ωzp = . (19)
L
experimental transient responses of vO due to a step change
in output current of ∆io = 0.1 A, respectively, and were in In (17), ωzn is given by (13), ωn is given by (14), and ζ is given
excellent agreement. by (15). In experimental verification, there will be a delay Td
BRYANT AND KAZIMIERCZUK: VOLTAGE-LOOP POWER-STAGE TRANSFER FUNCTIONS WITH MOSFET DELAY 351

Fig. 15. Experimental step response of vO due to ∆D = 0.06. Upper trace is

step driving dT ; lower trace is vO . Horizontal scale is 1 ms/div. Vertical scale
for trace 1 is 100 mV/div.; vertical scale for trace 2 is 1 V/div.
Fig. 13. Tp Bode plots from Hewlett-Packard 4194A Impedance/Gain-Phase
Analyzer.

Fig. 16. Small-signal model to determine Mv for boost converter operating

in CCM.

Fig. 14. Step response of vO due to ∆D = 0.06. For the example boost converter from Section III, we ob-
tain Tpx = −0.2283 V, Tp (0) = 33.902 V = 30.6 dBV, fzp =
from the spectrum-analyzer oscillator to the duty cycle due to 7.497 kHz, and the 3-dB bandwidth f−3 dB is approximately
the duty-cycle modulator, MOSFET driver, and the MOSFET 800 Hz. Tp is a nonminimal phase transfer function. However,
itself. For high-power applications, the MOSFET delay will for this example, the zeros ωzn and ωzp cancel each other,
be dominant. The delay will therefore be primarily determined so that the phase does not approach −270◦ . The duty-cycle
by the power MOSFET used and should be experimentally modulator was a LM357N op-amp used as a comparator with
measured. The delay can be represented with the function e−sTd a reference ramp of 5 V, giving a duty-cycle-modulator gain
from dc to fs /2. A first-order Padé function can be used to of 20 log(1/5) = −14 dB. Theoretical and experimental Bode
approximate e−sTd as a rational function plots for Tp are shown in Figs. 11–13, respectively. The mea-
sured delay of the IRF520 power MOSFET was dominant and
1− Td s s− 2
was Td = 1 µs. A MOSFET driver was not used, and the op-
e−sTd ≈ 2
=− Td
. (20)
1+ Td s
2
s+ 2
Td
amp was very fast, having negligible delay. Figs. 11–13 show
excellent agreement. In Fig. 13, we must add 14 dB to the mag-
Using (20), (17) is given by nitude response to compensate for the duty-cycle-modulator-
gain loss. Figs. 14 and 15 show transient responses of vO due to
(s − ωzp )(s + ωzn ) s −
2
Td a step change in duty cycle ∆D = 0.06. The experimental and
Tp (s) = −Tpx (21)
s2 + 2ζωn s + ωn2 s + 2
Td
theoretical responses are in excellent agreement. In the experi-
mental response, shown in Fig. 15, trace 1 is the step function
and Tp (0) is for dT . ∆D is found by using the duty-cycle-modulator gain,
∆D = ∆VR /Vramppp = (0.3 V)/(5 V) = 0.06. In Fig. 15,
VO (1 − D)2 RL − r trace 2 is the output voltage. The peak response of vo
Tp (0) = . (22)
(1 − D) [(1 − D)2 RL + r] is 2.7 V.
352 IEEE TRANSACTIONS ON INDUSTRIAL ELECTRONICS, VOL. 54, NO. 1, FEBRUARY 2007

Fig. 17. Mv magnitude response. Fig. 20. Step response of vO due to ∆vi = 0.875 V.

Fig. 21. Experimental step response of vO due to ∆vi = 0.875. Upper trace
is ∆vi ; lower trace is ∆vo . Horizontal scale: 2 ms/div. Vertical scale for upper
trace is 500 mV/div.; vertical scale for lower trace is 500 mV/div.

V. I NPUT V OLTAGE TO O UTPUT V OLTAGE

Fig. 18. Mv phase response.
T RANSFER F UNCTION
The input voltage to output-voltage transfer function Mv can
be found by setting input d and disturbance input io of Fig. 3 to
zero, resulting in the circuit of Fig. 16. This gives
vo (s) s + ωzn
Mv (s) = |d=io =0 = Mvx 2 (23)
vi (s) s + 2ζωn s + ωn2
where
rC RL (1 − D) (1 − D)
Mvx = = (rC RL ) . (24)
L(rC + RL ) L

Mv (0) is

(1 − D)RL
Mv (0) = . (25)
(1 − D)2 RL + r
Also, ωzn is given by (13), ωn is given by (14), and ζ is given
by (15). For the example boost converter from Section III,
we obtain Mvx = 608.4 1/s, Mv (0) = 1.918 = 5.656 dB, and
Fig. 19. Mv Bode plots from Hewlett-Packard 4194A Impedance/Gain-Phase f−3 dB ≈ 900 Hz. The theoretical and experimental Bode plots
Analyzer. of Mv for this example are shown in Figs. 17–19, respectively,
BRYANT AND KAZIMIERCZUK: VOLTAGE-LOOP POWER-STAGE TRANSFER FUNCTIONS WITH MOSFET DELAY 353

and are in excellent agreement. Figs. 20 and 21 show theoretical [9] M. K. Kazimierczuk and D. Czarkowski, “Application of the principle
and experimental transient responses of vO due to a step change of energy conservation to modelling the PWM converters,” in Proc. 2nd
IEEE Conf. Control Appl., Sep. 13–16, 1993, pp. 291–296.
in input voltage of ∆vi = 0.875 V, respectively, and are in [10] S. Ben-Yaakov and D. Edry, “Averaged models and tools for studying
excellent agreement. the dynamics of switch mode dc–dc converters,” in Proc. IEEE Power
Electron. Spec. Conf., Taipei, Taiwan, R.O.C., 1994, pp. 1218–1225.
[11] B. Bryant and M. K. Kazimierczuk, “Small-signal duty cycle to inductor
VI. C ONCLUSION current transfer function for boost PWM dc–dc converter in continu-
ous conduction mode,” in Proc. IEEE ISCAS, Vancouver, BC, Canada,
The small-signal transfer functions from output current to May 23–26, 2004, pp. 856–859.
[12] ——, “Open-loop power-stage transfer functions relevant to current-mode
output voltage Zo , duty cycle to output voltage Tp , and input control of boost PWM converter operating in CCM,” IEEE Trans. Circuits
voltage to output voltage Mv have been derived for the power Syst. I, Reg. Papers, vol. 52, no. 10, pp. 2158–2164, Oct. 2005.
stage of a boost PWM dc–dc converter operating in CCM using
an averaged linear time-invariant circuit model of the nonlinear
converter [8], [9], which has a simpler topology than that
Brad Bryant (M’98) received the M.S. and Ph.D.
presented in [6]. Zo , Tp , and Mv are second-order functions, degrees in electrical engineering from Wright
neglecting the MOSFET delay for Tp . The MOSFET delay State University, Dayton, OH, in 2000 and 2004,
for Tp has been taken into account by using a first-order Padé respectively.
He was a Lab Teaching Assistant from 1998 to
approximation for the delay function e−sTd . Tp is a nonminimal 1999, and a Graduate Teaching Assistant from 1999
phase function. Tp phase response for the given example goes to to 2004, with the Department of Electrical Engi-
−180◦ due to the zero ωzn canceling out the phase of the RHP neering, Wright State University. His research inter-
ests are in high-frequency resonant and PWM dc/dc
zero ωzp , neglecting the delay. The delay of the MOSFET for Tp power converters, dc/ac inverters, high-frequency
in the given example circuit starts to have an effect at approxi- rectifiers, modeling and control of converters, hi-
mately fs /20 and adds approximately 20◦ of phase shift (delay) fidelity vacuum tube audio equipment, and power semiconductor devices.
Dr. Bryant was named IEEE Wright State Student of the Year 1999, WSU
at fs /2. The MOSFET delay can have effect on high-frequency Electrical Engineering Student of the Year 1999, and WSU Electrical Engi-
performance and loop stability when wide bandwidth con- neering Graduate Student of the Year 2000 and 2004. He is a member of Tau
trollers are used for both the current and voltage control loops. Beta Pi.
The transfer functions Zo , Tp , and Mv are dependent upon con-
verter topology and mode of operation (CCM/discontinuous-
conduction mode). Zo is important to determine the closed-loop Marian K. Kazimierczuk (M’91–SM’91–F’05)
output impedance in both current-mode and voltage-mode con- received the M.S., Ph.D., and D.Sci. degrees in elec-
tronics engineering from the Department of Elec-
trol of PWM converters. Tp is important to describe the voltage tronics, Technical University of Warsaw, Warsaw,
loop for current-mode control and is essential in voltage-mode Poland, in 1971, 1978, and 1984, respectively.
control. Mv is important to determine the closed-loop audio He was a Teaching and Research Assistant from
1972 to 1978 and Assistant Professor from 1978 to
susceptibility in both current-mode and voltage-mode control 1984 with the Department of Electronics, Institute of
of PWM converters. The theoretical and experimental results Radio Electronics, Technical University of Warsaw,
were in excellent agreement, confirming the validity of all the Poland. In 1984, he was a Project Engineer
for Design Automation, Inc., Lexington, MA. In
transfer functions derived. The transfer functions are valid from 1984–1985, he was a Visiting Professor with the Department of Electrical
dc to half of the switching frequency and can be used to model Engineering, Virginia Polytechnic Institute and State University, Blacksburg.
the voltage loop of PWM converters. The methodology used Since 1985, he has been with the Department of Electrical Engineering,
Wright State University, Dayton, OH, where he is currently a Professor. His
can be applied to other converter topologies. research interests are in high-frequency high-efficiency switching-mode tuned
power amplifiers, resonant and PWM dc/dc power converters, dc/ac inverters,
high-frequency rectifiers, modeling and control of converters, high-frequency
R EFERENCES magnetics, power semiconductor devices, electronic ballasts, RF transmitters,
[1] G. W. Wester and R. D. Middlebrook, “Low-frequency characterization sensors, and superconductivity. He is the coauthor of the books Resonant Power
of switched dc-to-dc converters,” IEEE Trans. Aerosp. Electron. Syst., Converters (Wiley, 1995), Electronic Devices, A Design Approach (Prentice-
vol. AES-9, no. 5, pp. 376–385, May 1973. Hall, 2004), and Lab Manual (Prentice-Hall, 2004). He has published over 250
[2] R. D. Middlebrook and S. Ćuk, “A general unified approach to modeling technical papers, 95 of which have appeared in IEEE TRANSACTIONS AND
switching converter power stages,” in Proc. IEEE Power Electron. Spec. JOURNALS. He is the holder of eight patents.
Conf., 1976, pp. 18–34. Prof. Kazimierczuk received the IEEE Harrell V. Noble Award for his
[3] ——, Advances in Switched-Mode Power Conversion, 2nd ed, vol. I/II. contributions to the fields of aerospace, industrial, and power electronics in
Pasadena, CA: TESLAco, 1983. 1991. He is also a recipient of the 1991 Presidential Award for Faculty Ex-
[4] W. M. Polivka, P. R. K. Chetty, and R. D. Middlebrook, “State-space aver- cellence in Research, the 1993 and 2000 College of Engineering and Computer
age modeling of converters with parasitics and storage time modulation,” Science Teaching Awards, the 1995 Presidential Award for Outstanding Faculty
in Proc. IEEE Power Electron. Spec. Conf., 1986, pp. 119–143. Member, the 1996–1999 Brage Golding Distinguished Professor of Research
[5] R. P. Severns and G. Bloom, Modern dc-to-dc Switchmode Power Con- Award from Wright State University, the 1997 and 2002 College Professional
verter Circuits. New York: Van Nostrand, 1985. pp. 30–42 and 130–135. Service Awards, and the 2004 Wright State University Board of Trustees’
[6] V. Vorpérian, “Simplified analysis of PWM converters using model of Award. He was an Associate Editor of the IEEE TRANSACTIONS ON CIRCUITS
PWM switch. Part I: Continuous-conduction mode,” IEEE Trans. Aerosp. AND S YSTEMS I, in 1993–1995, 1997–2001, 2002–2003, and 2004–2005,
Electron. Syst., vol. 26, no. 3, pp. 490–496, May 1990. and served as an Associate Editor for the Journal of Circuits, Systems, and
[7] ——, “Simplified analysis of PWM converters using model of PWM Computers. He serves as an Associate Editor for the IEEE TRANSACTIONS
switch. Part I: Discontinuous-conduction mode,” IEEE Trans. Aerosp. ON I NDUSTRIAL E LECTRONICS and the IEEE T RANSACTIONS ON C IRCUITS
Electron. Syst., vol. 26, no. 3, pp. 497–505, May 1990. AND S YSTEMS I. He was a member of the Superconductivity Committee of
[8] D. Czarkowski and M. K. Kazimierczuk, “Energy-conservation approach the IEEE Power Electronics Society. He was a Chair of the CAS Technical
to modeling PWM dc–dc converters,” IEEE Trans. Aerosp. Electron. Syst., Committee of Power Systems and Power Electronics Circuits in 2001–2002
vol. 29, no. 3, pp. 1059–1063, Jul. 1993. and 2004–2005. He is a member of Tau Beta Pi.