You are on page 1of 12

This article has been accepted for publication in a future issue of this journal, but has not been

fully edited. Content may change prior to final publication. Citation information: DOI 10.1109/TPEL.2017.2716830, IEEE
Transactions on Power Electronics

Small-Signal Model of Flyback Converter in


Continuous-Conduction Mode with Peak-Current
Control at Variable Switching Frequency
Ching-Hsiang Cheng1, Ching-Jan Chen1, Member, IEEE, Shinn-Syyong Wang2
1
Department of Electrical Engineering, National Taiwan University, Taipei, Taiwan
2
Richtek Technology Corporation, Chupei, Taiwan

Abstract—Flyback converter is widely used in low-power Figs. 1 and 2 show a practical frequency characteristic
adaptor applications. To achieve high efficiency for the entire of frequency reduction control and the simplified circuit
load range of operations, the variable-frequency peak-current- diagram of VFPCM control for flyback converter,
mode (VFPCM) control is often used in flyback converter to respectively. Fig. 1 shows that the switching frequency, fs,
reduce switching frequency at light-load. Since there is no small- decreases when the compensated voltage VC decreases. Since
signal model being developed for the VFPCM control to VC is equal to the sensed peak switch current and is
overcome the unstable problem in a wide range of operations. proportional to load current, the switching frequency is
In the present study, a novel, accurate small-signal model for reduced when the load current is reduced. In practical
VFPCM control operated in continuous-conduction mode applications, maximum and minimum switching frequency
(CCM) is proposed. Control behavior and design are analyzed limitations are generally used to avoid EMI effect and audio
and compared with constant-frequency peak-current-mode noise, respectively. The converter control is set to constant-
(CFPCM) control to give engineers insight. A control IC and a frequency peak-current-mode (CFPCM) when the switching
flyback converter are built for experimental validation of the frequency reaches maximum or minimum value.
proposed model. Experimental and simulation results verify the
proposed model. fs
CFPCM VFPCM CFPCM
Index Terms—Flyback converter, variable-frequency
fmax
peak-current-mode control (VFPCM), small-signal modeling,
describing function (DF)

I. INTRODUCTION
fmin
F lyback converter is widely used in low-power adapter
applications due to its cost effectiveness and electrical
isolation characteristics [1]. For example, flyback converter
VC1 VC2 VC

is often adopted to supply below 100W off-line power for Fig. 1 Practical frequency characteristics of frequency reduction control
cellphone, notebook, and personal computer. In recent years, ID Vo
flyback converter with high efficiency over the entire load Lm RCo
n 1
range has become increasingly important to respond the Vin Co RL IO
demands for energy efficiency and specifications [2]. Vac
Variable-frequency peak-current-mode (VFPCM) RCS
control is a common control method to obtain high efficiency Se
Vreg Rc3
Driver VCS
over the entire load range [3-8]. This control is also called Rd
Q R VC Ra
frequency reduction control. The controller adapts its d Q S
Cb Rca Ca
switching frequency according to the load conditions. At Trigger IA ITon Opto-
coupler
heavy-load condition, the controller keeps higher switching Signal Kgen
TL431 Rb
frequency to reduce switching ripple and ripple-related VTon

conduction loss. When the load decreases, the controller Vth CTon Compensator

reduces its switching frequency to reduce switching loss and


Controller
maintain light-load efficiency.
Fig. 2 Simplified circuit diagram of VFPCM control

This work was supported by a research grant from Richtek Corporation Although the VFPCM control is widely used in practical
to Taiwan University, both in Taiwan. applications, no complete small-signal model for VFPCM
C.-H. Cheng and C.-J. Chen are with the Department of Electrical control has been reported in literature, especially in
Engineering, National Taiwan University, Taipei 10617, Taiwan, R.O.C. continuous-conduction mode (CCM) [9, 10]. The
(wind19871219@gmail.com, corresponding author: Ching-Jan Chen. Email:
chenjim@ntu.edu.tw Phone: +886-2-33663550. Postal address:EE2-348,
compensation design for VFPCM controlled flyback
No.1, Sec. 4, Roosevelt Road, Taipei 10617, Taiwan, R.O.C.) converter is usually based on trial-and-error approach or the
S.-S. Wang is with the Richtek Technology Corporation, Chupei City, model of CFPCM [11], which apparently differs from
Hsinchu, Taiwan, R.O.C VFPCM in small-signal behavior. This results in possible

0885-8993 (c) 2016 IEEE. Personal use is permitted, but republication/redistribution requires IEEE permission. See http://www.ieee.org/publications_standards/publications/rights/index.html for more information.
This article has been accepted for publication in a future issue of this journal, but has not been fully edited. Content may change prior to final publication. Citation information: DOI 10.1109/TPEL.2017.2716830, IEEE
Transactions on Power Electronics

unstable problem, tedious control loop design, and lack of engineers insight into the small-signal behaviors of each part
tool to optimize control of the system. of converter. Based on the model, the compensation can be
Below shows a near unstable example of VFPCM designed for loop stability and faster transient response. The
controlled flyback converter when compensation is designed model of VFPCM in discontinuous conduction mode (DCM)
based on the small-signal model of CFPCM. Figs. 3 and 4 are is also derived, but it will not be discussed in this paper due
the measured time domain waveforms of a 20W (5V/4A) to its more stable characteristics than CCM.
charger operated in CFPCM and VFPCM, respectively. Both The paper is organized as follows. The description and
circuits are operated in CCM. As shown in Fig. 3, the steady-state frequency reduction analysis of VFPCM
converter is stable when Io is equal to 4A, where the control controlled flyback converter are presented in section II. The
is set to CFPCM to clamp switching frequency to a maximum proposed small-signal model for VFPCM in CCM is
value. When Io decreases to 3A, the control enters into presented in section III. Section IV compares the
VFPCM. Fig. 4 shows that the operation of converter performance between VFPCM and CFPCM. Some
becomes near unstable and serious jittering phenomenon comments with engineer insight are also summarized.
occurred. Hence, it is indispensable to develop a small-signal Experimental verifications are demonstrated in Section V.
model for VFPCM controlled flyback converter. This is Finally, conclusions are made in Section VI.
especially true for the recently proposed universal serial bus
power delivery (USB PD) applications [12]. The unstable II. DESCRIPTION AND STEADY-STATE ANALYSIS OF
problem is often to appear in cellphone and notebook VFPCM CONTROL
applications, where the output voltages are respectively 5V
Fig. 2 shows the simplified circuit diagram of VFPCM
and 20V, due to wide range of operation conditions of USB
controlled flyback converter. The AC power passes through
PD.
a full bridge diode rectifier, an input capacitor, and a flyback
VGS converter to generate a regulated DC output voltage V O for
load resistor RL. Lm is the magnetizing inductor of
VC transformer. ID is the current of diode. The output capacitor
VCS is represented as Co in series with equivalent series resistor
VO
(ESR) RCo. The output voltage is sensed through an isolated
type-II compensator consisting of an opto-coupler and a
TL431 to generate compensated voltage VC. In the controller,
the compensated voltage VC, the sensed current signal VCS
VGS: 7V / div and the slope compensation Se determine the on-time.
VC : 80mV / div (offset: 1.38V) Comparison with traditional CFPCM, the compensated
VCS: 200mV / div voltage VC of VFPCM determines not only on-time but also
VO : 200mV / div (offset: 5V) switching frequency.
Time scale: 5uS / div
Fig. 5 shows the steady-state waveforms of VFPCM in
Fig. 3 Time domain waveform of CFPCM (Io =4A) CCM. When the voltage VTon of capacitor CTon is charged to
Vth by the bias current IA plus ITon, a new on-time starts. Since
VGS ITon is proportional to VC with the gain of Kgen, switching
frequency fS is determined by VC. The duty turn-off instant is
VC
determined by the sensed inductor current iL when it reaches
VCS the VC minus slope compensation Se, which is the same as
peak current-mode control. Therefore, when load current IO
VO
decreases, VC and ITon also decrease. This results in reduced
switching frequency and increased light load efficiency for
VFPCM control.
The derivation of switching frequency versus load
VGS: 7V / div
VC : 80mV / div (offset: 1.38V) current characteristics for VFPCM control is important for
VCS: 200mV / div efficiency optimization and control IC parameter design. In
VO : 200mV / div (offset: 5V) the traditional CFPCM, switching frequency is easy to define.
Time scale: 5uS / div In VFPCM, however, this definition becomes more
Fig. 4 Time domain waveform of VFPCM (Io =3A) complicate due to the load dependent characteristics of
switching frequency. Fig. 5 shows that the average diode
Quasi-resonant (QR) control is another common method current ID, which equals to output current IO at steady state,
to obtain high efficiency [13]. QR control adopts valley can be obtained from equation (1). Here 100% converter
switching to reduce switching loss. However, in order to keep efficiency is assumed for simplicity. The relation between VC
valley switching, the QR control may force switching and switching period TSW is represented by equation (2)
frequency to increase at light load, but results in poor light according to capacitor charge equation. The relationship
load efficiency [14]. Currently, VFPCM and QR control are between rising slope Sn and falling slope Sf of the sensed
both commonly used for high efficiency adapter applications. inductor current can be described by equation (3) based on
In this paper, a small-signal model of VFPCM voltage-second-balance principle. Combining equations (1),
controlled flyback converter in CCM is proposed to solve the (2), and (3), the bias current IA can be obtained from equation
above mentioned problems. The proposed model also gives (4). Based on equation (4), the IA and Kgen can be used to

0885-8993 (c) 2016 IEEE. Personal use is permitted, but republication/redistribution requires IEEE permission. See http://www.ieee.org/publications_standards/publications/rights/index.html for more information.
This article has been accepted for publication in a future issue of this journal, but has not been fully edited. Content may change prior to final publication. Citation information: DOI 10.1109/TPEL.2017.2716830, IEEE
Transactions on Power Electronics

design the frequency reduction function at different IO Level. constant value and the control operates in CFPCM. Therefore,
An appropriate design for the frequency reduction function the four regions can be classified as VFPCM in CCM or
can optimize light load efficiency according to different DCM, and CFPCM in CCM or DCM.
working conditions of converter. Fig. 6 illustrates that a Before obtaining the relation between switching
converter is operated at specific input voltage, output voltage, frequency and output current, the boundary condition of
and slope compensation. The switching frequency of VFPCM between CCM and DCM needs to be determined
converter at IO2 and IO3 can be designed according to IA and first. According to the definition of inductor current in
Kgen of equation (4). Therefore, the frequency reduction boundary condition, the relation between compensated
curve can be determined directly. voltage and inductor current slope can be described by
n 1  D  equation (5). By substituting equations (5) and (2) into
IO  I D   2 VC  Se DTS   S f 1  D  TS  (1) equation (1), the boundary condition of output current can be
2R CS
determined from equation (6).
TSW  I A  K genVC   VthCTon (2) VC _ Boundary   S n  Se  DTS  0 (5)
1
S n DTS  S f 1  D  TS (3) n 2 1  D  VO  I A
2
 I A  K gen  Sn  Se  D 
2

I O _ Boundary      (6)
 2VthCTon 
R IO  2 Se  S n  D  2 Lm
  2VthCTon  VthCTon

I A  Vth CTon f S  K gen  CS   (4)
 n 1  D  2 fS  When VFPCM is operated in CCM, the switching
frequency, fS, can be derived by substituting equation (2) into
where
equation (1). The result is shown in equation (7), which can
Vin nV be used to obtain the relationship between fS and IO as shown
Sn  RCS  , S f  RCS  O
Lm Lm in Fig. 6.
 
2n I A  VthCTon 8n  2Se +Sn  D 
2
K gen R CS  2 I O 2n I A   2IO 2

fS  
  
      (7)
4VthCTon  1  D  R CS K  
 1  D  R CS K gen  K gen R CS 2 
 
iD
ID
When VFPCM is operated in DCM, the derivation of
(1-D)Ts switching frequency is similar to that in CCM.
Vc Se
Sn
RCS iL RCS IL III. PROPOSED SMALL-SIGNAL MODEL OF VFPCM
Sf
CONTROL
Vth The modeling strategy of VFPCM control is illustrated
Sr
VTon in Fig. 7. Since VFPCM control contains highly nonlinear
signal of inductor current and variable frequency operation,
Trigger
Signal the describing function approach [15] is used to accurately
Ts=1/fs model the nonlinear part of the control. The nonlinear part of
d DTs the control such as inductor, switch, and modulation circuit
t1 t2 t3 are modeled as a single entity blocked by Factor1 as shown
Fig. 5 Steady-state waveforms of VFPCM control in CCM in Fig. 7. On the other hand, the conventional average model
[19], which ignores side-band and switching frequency
fs CFPCM VFPCM VFPCM CFPCM information, cannot predict high-frequency response of
in DCM in DCM in CCM in CCM
control loop. However, it is accurate enough to predict power
fmax
stage such as the circuits blocked by Factor2 and Factor3.
Therefore, these two parts are modeled by the average model.
The proposed modeling approach provides engineers insight
fmin into the small-signal behaviors of each part of control.
fs VC1 VC2 VC3 VC The control to output transfer function Gvc(s), defined as
fmax vo/vc, can be expressed by three main transfer functions,
namely Factor1(s), Factor2(s), and Factor3(s) in
mathematical derivation. As seen in Figs. 7 and 8, the transfer
functions shown in Fig. 8 can describe the small-signal
fmin
behavior of corresponding parts of the circuit in Fig. 7.
IO1 IO2 IO3 IO Factor1(s), defined as iL/vc, is derived by treating the inductor,
Fig. 6 Frequency versus VC and IO of practical frequency control modulator, current control loop, and switches as a single
entity. Factor2(s), defined as iD/iL, is the current transfer
The operation modes of the flyback control, as shown in function from inductor current to diode current. The
Fig. 6, can be divided into four regions. The VC2 is the Factor3(s), defined as vo/iD, is a network transfer function
boundary condition for CCM and DCM. So the control consisted of output capacitor and load resistor. The transfer
operates in CCM when VC is larger than VC2. In contrary, the function k2, defined as iD/vo, is to describe the effect of output
control operates in DCM when VC is smaller than VC2. The voltage perturbation on diode average current as VC keeps
control operates in VFPCM when VC is between VC1 and VC3. constant. Based on the result of transfer function G vc(s), the
For the other VC regions, the switching frequency is fixed at

0885-8993 (c) 2016 IEEE. Personal use is permitted, but republication/redistribution requires IEEE permission. See http://www.ieee.org/publications_standards/publications/rights/index.html for more information.
This article has been accepted for publication in a future issue of this journal, but has not been fully edited. Content may change prior to final publication. Citation information: DOI 10.1109/TPEL.2017.2716830, IEEE
Transactions on Power Electronics

stability criterion of current-loop and voltage-loop The assumptions made in the derivation are summarized
compensation design can be established. as follows:
1. The magnitude of the inductor-current slopes remains
ID Vo constant during the on-period and off-period;
Lm
2. The magnitude of the perturbation signal is very small
n 1 RCo
as compared to DC value;
Vin IL RL IO
Co 3. The perturbation frequency fm and the switching
Factor2 Factor3 frequency fS are commensurable, which means that
N×fS = M×fm, where N and M are positive integers;
Driver Critical derivation steps are listed below:
VCS Se Step 1: Perturbed on-time and off-time calculations
RCS
based on Fig. 9 are represented by equations (8) and (9):
Q R Vc Vth CTon
d I A  K gen vc (ti  TON (i )  TOFF (i ) )  (8)
Q S
Trigger TON (i )  TOFF (i )
fm
Signal
IA ITon
Kgen vc (ti 1  TON (i 1) )  SeTON (i 1)  S f TOFF (i 1)  vc (ti  TON (i ) )  SeTON (i )  SnTON (i ) (9)
VTon where
CTon Vin nV
Vth Sn  RCS  , S f  RCS  O
Factor1
Lm Lm
Step 2: Perturbed duty cycle and inductor current
calculations are obtained from equations (10) and (11).
Fig. 7 Modeling strategy of VFPCM control M
d t  0  t  t TON ( M ) TOFF ( M )
= u (t  ti )  u (t  ti  TON (i ) )  (10)
M
i =1
Vin(s)

 V d (t )  nV 1  d (t ) dt  i
1
iL  t  0 t  t
t

TON ( M ) TOFF ( M )
= in O L0 (11)
k1 M Lm 0

where
iL(s) iD(s) Vo(s) i 1
VC(s)
Factor1(s) Factor2(s) Factor3(s) ti  (i  1)(TON  TOFF )    T
k 1
ON ( k )  TOFF ( k ) 
k2 Step 3: Fourier analysis of inductor current is performed
GVC(s)
to obtain the describing function result, as shown in (12).
Fig. 8 Control block diagram of VFPCM control
iL ( s)  K genTSW 1  e sTON S f e sTON 
2

   sTSW
  sTSW 
A. Derivation of transfer functions Factor1(s) of VFPCM vc ( s)  VthCTon  1  e ( S e  S n )  ( S e  S f ) e 
control in CCM (12)
1  e sTSW  VIN  nVO
 
The transfer function of Factor1(s), which is vc to iL ( Se  Sn )  ( Se  S f )e sTSW  sLmTSW
transfer function, can be derived by using the describing
After applied Padé approximation and neglect high
function approach [15]. The steady-state waveforms and
frequency effect, equation (12), Factor1 transfer function, can
perturbed waveforms are shown in Fig. 9, where the dotted
be simplified to equation (13).
and solid lines denote waveforms at steady state and
perturbation, respectively. When VC is perturbed, the on-time 1
i ( s )  DK genTSW  2Se  S n  
2
RCS
changes. Moreover, the frequency also changes because the Factor1( s )  L   1 (13)
vc ( s )  2VthCTon 
1 s s2
rising slope Sr of Vton changes. 
Q11 1
2

vc(t) Se where
Sn
RCS iL(t) 1 
Q1  , 1 
Sf

Vth 1 ( Se  S f )  Tsw
  
 Sn  S f  
Sr
VTon 2

Trigger
Signal
B. Derivation of transfer functions Gvc(s) of VFPCM control in
PWM
Toff(1) Toff(2) CCM
PWM
Ton(1) Ton(2) According to Fig. 8, the control to output transfer
t1 t2 t3 t4 ti tM
function Gvc(s) can be derived as shown in equation (14).
Fig. 9 Steady-state and perturbed waveforms for VFPCM control in CCM

0885-8993 (c) 2016 IEEE. Personal use is permitted, but republication/redistribution requires IEEE permission. See http://www.ieee.org/publications_standards/publications/rights/index.html for more information.
This article has been accepted for publication in a future issue of this journal, but has not been fully edited. Content may change prior to final publication. Citation information: DOI 10.1109/TPEL.2017.2716830, IEEE
Transactions on Power Electronics

vO ( s) Factor 3  s  1  k2 RL 
Gvc ( s)   Factor1 s   Factor 2  s   (14) 1
vc (s) 1  k2 Factor 3  s   P1 Co  RCo  RL   RL RCo 
 1  k2 
The Factor2(s) can be derived using average model  RCo  RL 
approach [19]. Based on the inductor and diode current
waveforms at steady-state, the average diode current can be 1
 ESR
represented by equation (15). By perturbing the variables and RCo Co
expressing the small-signal variation in frequency domain,
the small-signal variation of the diode current is obtained (1  D ) 2 n 2 RL
RHP
from equation (16). According to steady-state-average DLm
modeling results, the duty to inductor current transfer
function is represented by equation (17). Substituting 1
equations (15) and (16) into equation (17), the Factor2(s) Q1 1 ( Se  S f ) 
  
 Sn  S f  
transfer function is obtained from equation (18).
2
I D  n(1  D) I L (15) 

iD ( s )  n(1  D )iL ( s )  nI L d ( s ) (16) 1  / Tsw

iL ( s) Vin  nVO nVO


 (17) D
Vin  nVO
d ( s) sLm

n 2 1  D  TSW n 1  D   
3
iD ( s ) DVin  DLm   
VO

n
S 1  D  TSW 
Factor 2( s )   1  2 s k2 
iL ( s ) VO  n (1  D) RL
2

(18) 2 Lm Vin  nVO   1  D  RL RCS e 

The Factor3(s) transfer function is the output network


impendence consisted of output capacitor and load resistor as C. Explanation of modeling results of Gvc,spf (s) of VFPCM
expressed in equation (19). control

vO ( s) RL 1  sCo RCo  According to the above derivations and using the


Factor 3( s)   (19)
iD ( s) 1  sCo  RCo  RL  parameters listed in Table I, the Gvc,spf(s) of VFPCM control,
which can be illustrated in Bode plot as shown in Fig. 10, is
The k2 derivation also uses average modeling approach. a system with three poles and two zeros. There is a dominant
By perturbing the variables and expressing the small-signal pole at low frequency, which relates mainly to load resistance
variation in frequency domain for equations (1) and (3), the when the other circuit parameters are fixed. A pair of
small-signal variations of the output voltage to diode current complex poles locates at half of the switching frequency,
and duty to output voltage are obtained from equations (20) where Q factor is determined by the duty cycle and the slope
and (21). Combining equations (1), (20), and (21), the k2 is compensation. This implies that the slope compensation
represented by equation (22). design for current loop stability is the same as CFPCM. There
 n n2 1  D  TSW  ˆ n2 1  D  TSW
2
n is a left-half-plane zero, the location of which is only related
iˆD    VC  TSW Se 1  2 D   VO  d  vˆO (20)
 RCS RCS Lm  2Lm to characteristics of output capacitor. And a right-half-plane
zero, the location of which is related to the duty cycle and
Vin +nVO  dˆ  n 1  D  vˆ O (21) power stage parameters.
iˆD n2 1  D  TSW n 1  D   VO  Gain(dB)
3
n
k2      Se 1  D  TSW  (22)
vˆO 2 Lm  in O     L CS
V  nV 1  D R R  fP1(VFPCM) VFPCM
Finally, using the expression for Factor1(s) from
equation (13), Factor2(s) from equation (18), Factor3(s) from
equation (19) and k2 from equation (22), equation (14) can be f1(VFPCM)
simplified to equation (23), which is defined as Gvc,spf (s). fESR(VFPCM) fSW
 s  s  f(Hz)
1  1  
v ( s)   ESR    RHP 
GVC , spf ( s )  o  GVC _ DC (23) fRHP(VFPCM)
vc ( s)  s  s s2 
1  1   2
 P1  Q11 1  Fig. 10 Bode plot of Gvc,spf (s) for VFPCM control in CCM
where the parameters are summarized in Table I.
In the present study, the proposed model gives engineers
TABLE I insight into the small-signal behaviors of each part of
Transfer function parameters of VFPCM in CCM converter as shown in Fig. 11. Comparison of Figs 7 and 11,
the relation of small-signal behavior and circuit parameters
Item VFPCM in CCM can be understood quickly. Moreover, it is worth noting that
the parameters of frequency reduction function such as K gen,
 DK genTSW 2  2 Se  Sn   1 DVin RL CTon, and Vth only affect the DC gain of Gvc,spf (s) and do not
GVC _ DC   1
2V C  R V 1  k2 RL  affect the location of poles and zeros.
 th Ton  CS O

0885-8993 (c) 2016 IEEE. Personal use is permitted, but republication/redistribution requires IEEE permission. See http://www.ieee.org/publications_standards/publications/rights/index.html for more information.
This article has been accepted for publication in a future issue of this journal, but has not been fully edited. Content may change prior to final publication. Citation information: DOI 10.1109/TPEL.2017.2716830, IEEE
Transactions on Power Electronics

1 E. Model Verifications by Simulation


 K DT 2  2Se  S n   RCS
Factor1   gen SW  1
 
 1 s  s  The transfer functions of Gvc,spf (s) and Zo(s) are verified
2
 2VthCTon
 2  1
DC Gain1  Q11 1 
by SIMPLIS simulation tool. The simulation accuracy of
DC Gain2  
SIMPLIS is widely proved in many power electronics
 
DVin  s 
Factor 2  1
VO  n 2 (1  D) 2 RL    research papers [15-18]. The circuit parameters and working

 DLm 

RHP  s 
1 
 1


conditions are listed in Tables II and III. Figs. 14 and 15 show
 C R  ESR
Factor 3

RL  o Co  the comparison of model and SIMPLIS simulation results.
1  k2  Factor 3 1  k2 RL  1  s
DC Gain3 1  k2 RL  Both results are in excellent agreement for VFPCM control
 RR 
Co  RCo  RL  1  k2 L Co  P1 in various conditions. For instance, the model predicts the
iL(s) iD(s) Vo(s)  RCo  RL 
VC(s) gain peaking at half switching frequency well for the case of
Factor1(s) Factor2(s) Factor3(s)
VO=20V and IO=2.5A as shown in Fig.14.
k2
GVC(s)
TABLE II
Fig. 11 Small-signal behaviors of each part of converter. Circuit parameters of simulation in Vac=90Vrms

D. Derivation of transfer functions Zo(s) of VFPCM Item VFPCM in CCM


Input voltage 127V (Vac=90Vrms)
According to the control block diagram as shown in Fig.
8, the control of output transfer function G vc(s) can be Output voltage 5V 20V
expressed as equation (24). Based on equation (24), the Output current 3A 2.5A
equivalent circuit is shown in Fig. 12. Then, the output Switching frequency fS 80kHz 80kHz
impedance can be obtained by adding a perturbing current Magnetic inductor Lm 515uH
source at output as shown in Fig. 13. The open-loop output Transformer turn ratio n 6
impedance ZO(s) can be further simplified from equation (25)
Output capacitor Co 2000µF
to equation (26). Equation (26) shows that the low and high
ESR of capacitor RCo 18mΩ
frequency performances of Zo(s) are dominated by parallel
impedances of -1/k2 and RL as well as -1/k2 and RCo, IA= -3.025 uA IA= -5.545 uA
respectively. Se= 102V/mS Se= 51V/mS
Reference design
Factor 3  s  Kgen= 6.3uA/V, CTon = 10pF
vO ( s )   vc ( s )  k1vin ( s )  Factor1 s   Factor 2  s  
1  k2 Factor 3  s  Rcs = 0.54Ω, Vth = 2V
(24)
 1 
 vO ( s )   vc ( s )  k1vin ( s )  Factor1 s   Factor 2  s   / / Factor 3  s   TABLE III
 k2 
Circuit parameters of simulation in Vin=265Vrms
vo  1 
ZO (s)    / / Factor 3  s   (25) Item VFPCM in CCM
io  k2 
1
Input voltage 375V (Vac=265Vrms)
  1  
  / / RL   Output voltage 5V 20V
 1    k2  
Z O ( s )   / / RL  1  sCo RCo  1  sCo RCo (26) Output current 3A 4.5A
  1 
 k2  
  / / RL / / RCo   Switching frequency fS 60kHz 98kHz
  k2 
Magnetic inductor Lm 515uH
Vo Factor3(s) Transformer turn ratio n 6
Output capacitor Co 2000µF
Factor1(s) RCo
ESR of capacitor RCo 18mΩ
vˆcg vˆcg -1/k2 RL
k1vˆin vˆc Factor2(s) Co IA= -3.025 uA IA= -5.545 uA
Se= 102V/mS Se= 51V/mS
Reference design
Kgen= 6.3uA/V, CTon = 10pF
Fig. 12 Equivalent circuit of VFPCM control Rcs = 0.54Ω, Vth = 2V

Vo Factor3(s) Gvc,spf (s) Zo(s)


60 20
vo 40
Vo=5V, Io=3A Vo=5V, Io=3A

RCo io 0
Gain(dB)

Gain(dB)

20

-1/k2 RL 0  20

Co  20
 40
 40 Model Model
Simulation Simulation
 60  60
3 4 5
1 10 100 110 110 110 1 10 100 110
3 4
110
5
110
Frequency(Hz) Frequency(Hz)
Fig. 13 Output impedance derivation of VFPCM control
Gvc,spf (s) Zo(s)
90 45
Vo=5V, Io=3A Vo=5V, Io=3A
45
Phase(degree)

Phase(degree)

0
0

 45  45

 90

0885-8993 (c) 2016 IEEE. Personal use is permitted, but republication/redistribution requires IEEE permission. See http://www.ieee.org/publications_standards/publications/rights/index.html
135 Model
 90
Model
for more information.
Simulation Simulation
 20

Gai

Gai
Ga

Ga
This article has been
 20 accepted for publication in a future issue of this journal, but has not been fully edited. Content may
 20
change prior to final publication. Citation information: DOI 10.1109/TPEL.2017.2716830, IEEE
 40 Model
 40
Model Transactions on Power ElectronicsModel  40
Model  40
Simulation Simulation Simulation Simulation
 60  60  60  60
3 4 5
1 10 100 110 110 110 110
3 4
110
5
110 110
3
110
4
110
5 3 4 5
1 10 100 1 10 100 1 10 100 110 110 110
Frequency(Hz) Frequency(Hz) Frequency(Hz) Frequency(Hz)
Gvc,spf (s) Zo(s) Gvc,spf (s) Zo(s)
90 45 90 45
Vo=5V, Io=3A Vo=5V, Io=3A Vo=20V, Io=4.5A Vo=20V, Io=4.5A
45 45
Phase(degree)

Phase(degree)

Phase(degree)

Phase(degree)
0 0
0 0

 45  45  45  45

 90  90
 90  90
 135 Model Model  135 Model Model
Simulation Simulation Simulation Simulation
 180  135  180  135
3 4 5
1 10 100 110 110 110 110
3
110
4
110
5 110
3
110
4
110
5 3 4 5
1 10 100 1 10 100 1 10 100 110 110 110
Frequency(Hz) Frequency(Hz) Frequency(Hz) Frequency(Hz)

Gvc,spf (s) Zo(s) Fig. 15 Comparison of model and SIMPLIS simulation results for VFPCM
60 20
Vo=20V, Io=2.5A Vo=20V, Io=2.5A control in Vin=265Vrms
40
0
IV. COMMENTS AND COMPARISON OF VFPCM AND CFPCM
Gain(dB)

Gain(dB)

20

0  20 CONTROL BEHAVIOR
 20
 40
 40 Model Model According to the above modeling results of VFPCM, the
Simulation Simulation
 60  60
proposed model can be degenerated to conventional CFPCM
3 4 5 3 4 5
1 10 100 110 110 110 110 110 110
Frequency(Hz)
1 10 100
Frequency(Hz)
when Kgen is set to zero. This can be explained from circuit
Gvc,spf (s) Zo(s)
diagram of Fig. 2. When Kgen is zero, the Vc does not
90 45 influence the switching frequency and then, the control
Vo=20V, Io=2.5A Vo=20V, Io=2.5A
45 method becomes conventional CFPCM. Moreover, the
Phase(degree)

Phase(degree)

0
0 proposed model can be expanded to buck-boost converter by
 45  45 setting transformer turns ratio n to 1.
 90
 90
Based on the same simulation parameters of Tables II
 135 Model
Simulation
Model
Simulation
and III, the degenerated model can be validated by simulation
 180
1 10 100 110
3
110
4
110
5
 135
1 10 100 110
3
110
4
110
5
as shown in Figs. 16 and 17. Furthermore, the model results
Frequency(Hz) Frequency(Hz) are even more accurate than those reported in literature [11],
especially for Vin = 265Vrms.
Fig.14 Comparison of model and SIMPLIS simulation results for VFPCM
For the buck-boost converter validation, the current
control in Vin=90Vrms
sensing resistor is set to 1. This value was not included in the
Gvc,spf (s) Zo(s)
60 20 model as reported in literature [20]. Figs.18 and 19 show that
Vo=5V, Io=3A Vo=5V, Io=3A
40 the degenerated model results are in excellent agreement with
0
the simulation results. Moreover, the proposed model
Gain(dB)

20

0  20 predictions are more accurate than those reported in literature


[20].
    pwr    
 20 180 180
M ( i)  20 log G ( i) P ( i)  arg G ( i)M ( i)  20 log G ( i) P pwr( i)  arg Gpwr( i)  
 40 pwr pwr pwr pwr  Gain/dB pwr
 40 Model Model Gvc,spf_CFPCM
Gain/dB (s) 40 Gain/dB (s)
Gvc,spf_CFPCM
Simulation Simulation 40 40
 60  60 Vo=20V, Io=2.5A Vo=5V, Io=3A
3 4 5 3 4 5
1 10 100 110 110 110 1 10 100 110 110 110 20
Frequency(Hz) Frequency(Hz) 20 20
Gain(dB)

Gain(dB)

Gvc,spf (s) Zo(s) 0


90 45 0 0

Vo=5V, Io=3A Vo=5V, Io=3A


45
Proposed Model  20 Proposed Model
 20  20
Phase(degree)

Phase(degree)

0 Model [11]
Model [11]
0
Simulation Simulation
 40 40
40
 45  45 3 4 55
1 10 100 110
3
110
4
110
5 11 10
10 100
100 1110
10
3 11104
10 1110
10

 90 Frequency(Hz) Frequency(Hz)
 90
 135 Model Model GPhase / Deg(s)
vc,spf_CFPCM GPhase / Deg(s)
vc,spf_CFPCM
Simulation Simulation 0 0
 180  135 Vo=20V, Io=2.5A Vo=5V, Io=3A
3 4 5 3 4 5
1 10 100 110 110 110 1 10 100 110 110 110
 45  45
Phase(degree)

Phase(degree)

Frequency(Hz) Frequency(Hz)

 90  90
Gvc,spf (s) Zo(s)
60 20
Vo=20V, Io=4.5A Vo=20V, Io=4.5A  135 Proposed Model  135
Proposed Model
40 Model
Model [11] Model [11]
0 Simulation
Simulation Simulation
Gain(dB)

Gain(dB)

20
 180  180
3 4 5 3 4 5
1 10 100 110 110 110 1 10 100 110 110 110
0  20
Frequency(Hz) Frequency(Hz)
 20
 40
 40 Model Model Fig. 16 Comparison of models and simulation of flyback converter with
Simulation Simulation
 60
3 4 5
 60 CFPCM control in Vin=90Vrms
1 10 100 110 110 110 1 10 100 110
3
110
4
110
5

Frequency(Hz) Frequency(Hz)
Gvc,spf (s) Zo(s)
90 45
Vo=20V, Io=4.5A Vo=20V, Io=4.5A
45
Phase(degree)

Phase(degree)

0
0

 45  45

 90
 90
 135 Model Model
Simulation Simulation
 180  135
3 4 5
1 10 100 110 110 110 1 10 100 110
3
110
4
110
5

Frequency(Hz) Frequency(Hz)
0885-8993 (c) 2016 IEEE. Personal use is permitted, but republication/redistribution requires IEEE permission. See http://www.ieee.org/publications_standards/publications/rights/index.html for more information.
This article has been accepted for publication in a future issue of this journal, but has not been fully edited. Content may change prior to final publication. Citation information: DOI 10.1109/TPEL.2017.2716830, IEEE
Transactions on Power Electronics

       
180 180
Mpwr( i)  20 log Gpwr( i) Ppwr( i)  arg Gpwr( i)M pwr( i)  20 log Gpwr( i) Ppwr( i)  arg Gpwr( i) 
Gain/dB  Gain/dB
40 Gain/dB (s)
Gvc,spf_CFPCM 40 Gain/dB (s)
Gvc,spf_CFPCM A. Comparison of Gvc,spf (s) between VFPCM and CFPCM
40 40
Vo=20V, Io=5A Vo=5V, Io=3A
20 20 The control to output transfer function of CFPCM,
20 20
Gvc,spf_CFPCM(s), is described by equation (27). The only
Gain(dB)

Gain(dB)
0 0
0 0 difference of Gvc,spf (s) between VFPCM and CFPCM is the
 20 Proposed Model  20 Proposed Model
DC gains. All the other poles and zeros are the same for either
 20  20
Model [11]
Simulation
Model [11]
Simulation
VFPCM or CFPCM.
40
40  4040
3 44 55 3 44 55
1110 10 1110 1110 10 1110
 s  s 
3 3
11 10
10 100
100 10 1110 10 11 10
10 100
100 10 1110 10

1  1 
Frequency(Hz) Frequency(Hz)

 ESR    RHP 
 GVC _ DC _ CF 
Phase / Deg(s) v ( s)
Gvc,spf_CFPCM GPhase / Deg(s)
vc,spf_CFPCM
GVC , spf_CFPCM ( s)  o (27)
 s  s2 
0 0
vc ( s) s
Vo=20V, Io=5A Vo=5V, Io=3A
 1   1   2 
 45  45
 P1  Q11 1 
Phase(degree)

Phase(degree)

 90  90
where
 135
Proposed Model  135
Proposed Model
Model [11] Model [11] 1 DVin RL
GVC _ DC _ CF =
RCS VO 1 k2 RL 
Simulation Simulation
 180  180
3 4 5 3 4 5
1 10 100 110 110 110 1 10 100 110 110 110
Frequency(Hz) Frequency(Hz)
Fig. 20 compares the Gvc,spf (s) of VFPCM and CFCCM
Fig. 17 Comparison of models and simulation of flyback converter with in Bode plot. It can be seen that two transfer functions have
CFPCM control in Vin=265Vrms different DC gains while the locations of poles and zeros are
  Ppwr( i)  argGpwr( i) 
180 180
Ppwr( i)  arg Gpwr( i) 
Gain/dB
  the same. Therefore, if the compensator is designed based on
20 Gain/dB (s)
Gvc,spf_CFPCM Gain/dB (s)
Gvc,spf_CFPCM
20 20 CFPCM model, there will be discrepancy on cross-over
Vo=20V, Io=2.5A Vo=5V, Io=3A
0 frequency and phase margin when the converter operates at
0 0
VFPCM control.
Gain(dB)

Gain(dB)

 20
 20  20

 40
 40 Proposed Model  40
Proposed Model Gain(dB)
Model [20] Model [20]

 60
60
Simulation
 60
Simulation
fP1(VFPCM) CFPCM
33
10 10 4 4 5 5
10
11 10
10 100
100 1110 1110 1110 1 10 100 110
3
110
4
110
5
VFPCM
Frequency(Hz) Frequency(Hz)
GPhase / Deg(s)
vc,spf_CFPCM GPhase / Deg(s)
0 0
vc,spf_CFPCM
fP1(CFPCM) f1(VFPCM)
Vo=20V, Io=2.5A Vo=5V, Io=3A fESR(VFPCM) f
 45  45 SW
Phase(degree)

Phase(degree)

f(Hz)
 90  90
fESR(CFPCM) fRHP(VFPCM)
Proposed Model Proposed Model
 135
Model
Model [20]
 135
Model [20] f1(CFPCM)
 180
Simulation
Simulation
 180
Simulation fRHP(CFPCM)
3 4 5 3 4 5
1 10 100 110 110 110 1 10 100 110 110 110
Frequency(Hz) Frequency(Hz)
Fig. 20 Bode plot of Gvc(s) for VFPCM and CFPCM controls in CCM
Fig. 18 Comparison of models and simulation of buck-boot converter with

Ppwr( i)  argGpwr( i) 
180 CFPCM control in Vin=90Vrms
Ppwr( i)  argGpwr( i) 
180 From previous analysis, the Gvc,spf (s) of VFPCM is a
 Gain/dB 
20 Gain/dB (s)
Gvc,spf_CFPCM Gvc,spf_CFPCM
Gain/dB (s) system with three poles and two zeros. There are many ways
20
Vo=20V, Io=5A
20
Vo=5V, Io=3A
to design compensation circuits. Typically, a Type-II
0
0
0
compensator as shown in Fig. 2 (with one pole at origin, one
low-frequency zero, and one high-frequency pole) is the most
Gain(dB)

Gain(dB)

 20
 20  20
suitable one in this case.
 40
 40
Proposed Model
 40
Proposed Model Fig.21 shows the comparison of loop gain between
Model [20] Model [20]
Simulation Simulation VFPCM and CFPCM control in CCM while the
 60
60  60
11 10
10 100
100 10
110
1
33 10
1110
44
1110
5 5
10 1 10 100 110
3
110
4
110
5 compensators are the same. The simulation conditions and
Frequency(Hz) Frequency(Hz) compensator parameters are listed in Table IV. The circuit
GPhase / Deg(s)
vc,spf_CFPCM GPhase / Deg(s)
vc,spf_CFPCM parameters are the same as those for output voltage 5V
0 0
Vo=20V, Io=5A Vo=5V, Io=3A condition listed in Table II. Fig.21 indicates that there is a
 45 great difference between VFPCM and CFPCM controls. The
Phase(degree)

 45
Phase(degree)

 90  90
cross-over frequencies are 15kHz and 8kHz for VFPCM and
CFPCM controls, respectively. While the phase margins are
 135 Proposed Model Proposed Model
Model [20]
 135
Model [20] 10° and 30°, respectively. In other words, when the operation
Simulation
 180  180
Simulation
mode of converter changes from CFPCM to VFPCM, the
3 4 5 3 4 5
1 10 100 110 110 110 1 10 100 110 110 110 system response may increase but the phase margin decreases.
Frequency(Hz) Frequency(Hz)
Moreover, it may cause unstable problem.
Fig. 19 Comparison of models and simulation of buck-boot converter with
CFPCM control in Vin=265Vrms

0885-8993 (c) 2016 IEEE. Personal use is permitted, but republication/redistribution requires IEEE permission. See http://www.ieee.org/publications_standards/publications/rights/index.html for more information.
This article has been accepted for publication in a future issue of this journal, but has not been fully edited. Content may change prior to final publication. Citation information: DOI 10.1109/TPEL.2017.2716830, IEEE
Transactions on Power Electronics

5 5
T (s) T (s)
100 0 4
4

Io / A
Io / A
fC2=15kHz
80  45 3 3 2A-4A(2A/us)

Phase(degree)
fC1=8kHz 2
 90 2
Gain(dB)

60
Simulation(CFCCM)
40  135 1 1 Simulation(VFCCM)
PM 1=30° 5.18 5.18
20  180
5.14 5.14

Vo / V
Vo / V
Simulation(CFCCM) Simulation(CFCCM)
0 Simulation(VFCCM)  225 Simulation(VFCCM) 5.1 5.1
Type-II compensator Type-II compensator PM 2=10°
 20  270
5.06 5.06
3 4 5 3 4 5
110 110 110 110 110 110
1 10 100 1 10 100
5.02 5.02
Frequency(Hz) Frequency(Hz) 4.9 4.9 4.95 4.95 5 5 5.05 5.05 5.1 5.1 5.15 5.15

time/mSecs
time/mSecs 50uSecs/div
50uSecs/div
Fig.21 Comparison of loop gain for VFPCM and CFCCM controls in CCM Fig. 23 Comparison of load step response in CCM
with same compensation
C. Comments
TABLE IV
Circuit parameters of loop gain comparison 1. Comparison of the results of VFPCM and CFPCM and
their physical meanings
VFPCM CFPCM
Item
in CCM in CCM
According to comparison of the results of VFPCM and
CFPCM, the obvious difference is the DC gain of Gvc,spf (s).
Input voltage 127V (Vac=90Vrms)
This difference can be explained intuitively from time
Output voltage 5V domain waveforms. Fig. 5 shows that the average inductor
Output current 3A current can be expressed by equation (28). By perturbing the
Switching frequency fS 80kHz variables, the small-signal variation is expressed by equation
Ra=Rb = 68kΩ, Rca= 1kΩ (29). In the CFPCM, there is no variation at switching period,
Compensator design Rc3 = 200Ω, Rd=20kΩ so the small-signal variation of T̂SW is equal to zero. In the
Ca = 22nF, Cb = 11nF VFPCM, the switching period is derived as shown in
Opto-coupler current gain 65% equation (30). Finally, the ˆiL / v
ˆ C of VFPCM and CFPCM
is summarized in equation (31). The equation also indicates
B. Comparison of Output Impedances and Transient Response the DC gain difference of Gvc,spf (s) between VFPCM and
between VFPCM and CFPCM CFPCM.
1  Sn DTS 
Based on the small-signal model of CFPCM and IL   VC  Se DTS  2  (28)
equation (26), the open-loop output impedances Zo(s) are the R CS  
same for VFPCM and CFPCM, because they have same k2
1   2 Se  S n  D ˆ 
and Factor3(s). However, the close-loop output impedances iˆL   vˆC  TSW  (29)
Zoc(s), defined as Zo(s)/[1+T(s)], are totally different due to R CS  2 
the different loop gains T(s). Fig. 22 shows that the Zoc(s) of
K genTSW 2
VFPCM is smaller than that of CFPCM when the frequency ˆ 
T vˆC (30)
SW
is less than the cross-over frequency. The simulation Vth CTon
conditions are the same as those listed in Table IV. Thus, the
 iˆL 1
load step response of CFPCM is slower than that of VFPCM,    in CFCCM 
as shown in Fig. 23, when test conditions and compensator  vˆC R CS
designs are the same. ˆ  DK genTSW 2  2 Se  S n  
(31)
 iL  1 1    in VFCCM 
 vˆC R CS  2Vth CTon 
20
Zo(s)
 20
Zoc(s) 
The difference of DC gain affects the cross-over
0  40
frequency of loop gain. Therefore, the system has different
Gain(dB)

Gain(dB)

 20  60 transient response at load variation. When there is a variation


fC1=8kHz
fC2=15kHz on output voltage, the VFPCM has an extra frequency loop
 40  80
Simulation(CFPCM) Simulation(CFPCM) to response the variation except the original on-time
Simulation(VFPCM) Simulation(VFPCM)
 60  100
modulation loop. However, the CFPCM only depends on on-
3 4 5 3 4 5
110 110 110
1 10 100 1 10 100 110 110 110 time modulation loop. Hence, transient response of VFPCM
Frequency(Hz) Frequency(Hz)
is better than CFPCM.
Fig. 22 Comparison of close-loop output impedances for VFPCM and
2. Relation between frequency reduction control design and
CFCCM controls
Gvc,spf (s)

Equation (2) can be represented by equation (32) since


frequency and period are reciprocal. Therefore, the partial
differential of switching frequency with respect to VC is
given by equation (33), which represents the slope of
frequency reduction curve. Because the GVC_DC and equation

0885-8993 (c) 2016 IEEE. Personal use is permitted, but republication/redistribution requires IEEE permission. See http://www.ieee.org/publications_standards/publications/rights/index.html for more information.
This article has been accepted for publication in a future issue of this journal, but has not been fully edited. Content may change prior to final publication. Citation information: DOI 10.1109/TPEL.2017.2716830, IEEE
Transactions on Power Electronics

(33) are in proportion, the slope design of frequency TABLE V


reduction curve affects the GVC_DC. In addition, it also affects Circuit parameters of experiment in Vin=90Vrms
the transient response since the GVC_DC and cross-over
frequency of loop gain are also in proportion. Item VFPCM in CCM

fS 
I A +K genVC 
(32)
Input voltage
127V
(Vac=90Vrms)
127V
(Vac=90Vrms)
Vth CTon
Output voltage 5V 20V
f S K gen Output current 3A 3.6A
 (33)
VC Vth CTon Switching frequency fS 80kHz 65kHz
Magnetic inductor Lm 515uH
V. EXPERIMENTAL VERIFICATIONS Transformer turn ratio n 6
Output capacitor Co 2000µF (two 1000uF cap. in parallel)
In order to validate the small-signal model, a prototype
ESR of capacitor RCo 18mΩ
flyback converter and a controller with VFPCM control are
implemented in the present study. Figs. 24 and 25 show the Current sense resistor
540mΩ 338mΩ
control IC chip layout and the photo of prototype flyback Rcs
converter, respectively. Tables V and VI list circuit Kgen= 6.3uA/V Kgen= 6.3uA/V
parameters and working conditions for the experiments. The Se= 102V/mS Se= 51V/mS
frequency responses are measured using the Bode 100 Controller parameters IA= -3.025 uA IA= -5.545 uA
network analyzer and the range of frequency measurements CTon = 10pF CTon = 10pF
is conducted from 1 Hz to 100kHz. In experiments, the output Vth = 2V Vth = 2V
voltage is selected as 5V and 20V, which are considered for Ra=Rb = 68kΩ, Rca=1kΩ
adapter applications of smart phone and notebook. Compensator design Rc3 = 2kΩ, Rd=20kΩ
Figs. 26 and 27 show the comparison of measured and Ca = 22nF, Cb = 1nF
calculated frequency responses of G vc,spf (s). It can be seen Opto-coupler current
65%
that the results of proposed model are in excellent agreement gain
with the measurements except those beyond the half of
switching frequency. The discrepancy beyond half of TABLE VI
switching frequency is due to side-band effect of close-loop Circuit parameters of experiment in Vin=265Vrms
measurement.
Item VFPCM in CCM
375V 375V
comparator

Input voltage
(Vac=265Vrms) (Vac=265Vrms)
PWM

confidential
voltage and thermal
protection circuit

circuit logic and bias Output voltage 5V 20V


circuit
Output current 3A 3.8A
Switching frequency fS 64kHz 100kHz
variable
frequency Magnetic inductor Lm 515uH
circuit
current Transformer turn ratio
6
protection MOSFET n
circuit driver
Output capacitor Co 2000µF (two 1000uF cap. in parallel)
Fig. 24 Control IC chip layout ESR of capacitor RCo 18mΩ
Current sense resistor
540mΩ
Rcs
EMI Input Transformer
Filter Cap. Diode Kgen= 6.3uA/V Kgen= 6.3uA/V
Se= 102V/mS Se= 51V/mS
Controller parameters IA= -2.725 uA IA= -4.245 uA
CTon = 10pF CTon = 10pF
Output Vth = 2V Vth = 2V
Cap. Ra=Rb = 68kΩ, Rca=1kΩ
Compensator design Rc3 = 2kΩ, Rd=20kΩ
Ca = 22nF, Cb = 1nF
Full Bridge Power Controller Opto-coupler current
65%
Rectifier MOSFET gain

Fig. 25 . A prototype flyback converter

0885-8993 (c) 2016 IEEE. Personal use is permitted, but republication/redistribution requires IEEE permission. See http://www.ieee.org/publications_standards/publications/rights/index.html for more information.
This article has been accepted for publication in a future issue of this journal, but has not been fully edited. Content may change prior to final publication. Citation information: DOI 10.1109/TPEL.2017.2716830, IEEE
Transactions on Power Electronics

Gvc,spf (s) Gvc,spf (s)


ACKNOWLEDGMENTS
60 60

40 40 This work was supported by research grant 105-S-C29


from Richtek Corporation to National Taiwan University,
Gain(dB)

Gain(dB)
20 20

0 0 both in Taiwan. The authors would like to thank Tom Yang


 20  20 and other engineers from Richtek Corporation for giving
 40 Model(Vo=5V)  40 Model(Vo=20V) valuable comments. The authors also want to thank SIMPLIS
Experiment(VFPCM) Experiment(VFPCM)
 60
1 10 100 110
3
110
4
110
5
 60
1 10 100 110
3
110
4
110
5
Technologies Corporation, U.S.A, for providing SIMPLIS
Frequency(Hz) Frequency(Hz) simulation tool.
Gvc,spf (s) Gvc,spf (s)
90 90 REFERENCES
45 45
Phase(degree)

U. S. Padiyar, V. Kamath, “Design and implementation of a universal


Phase(degree)

0 0
[1]
input flyback converter,” in Proc. IEEE ICEEOT, 2016, pp. 3428-3433.
 45  45
[2] Y. Li and J. Zheng, “A low-cost adaptive multi-mode digital control
 90  90
solution maximizing AC/DC power supply efficiency,” in Proc. IEEE
 135 Model(Vo=5V)  135 Model(Vo=20V)
Experiment(VFPCM) Experiment(VFPCM)
Appl. Power Electron. Conf., 2010, pp. 349–354.
 180
3 4 5
 180
3 4 5
[3] NXP semiconductor. Green chip SMPS control IC, TEA1733A
1 10 100 110 110 110 1 10 100 110 110 110

Frequency(Hz) Frequency(Hz)
datasheet. NXP semiconductor document. (2013). [Online]. Available:
http://www.npx.com/
Fig. 26 Model verification by experiment in Vin=90Vrms [4] ON semiconductor. Fixed frequency current mode controller for
flyback converters, NCP1237 datasheet. ON semiconductor document.
(2011). [Online]. Available: http://www.onsemi.com/
Gvc,spf (s) Gvc,spf (s) [5] Fairchild semiconductor. Highly integrated green-mode PWM
60 60
controller, FAN6862 datasheet. Fairchild semiconductor document.
40 40
(2012). [Online]. Available: http://www.fairchildsemi.com/
Gain(dB)

Gain(dB)

20 20
[6] Leadtrend technology. Green-mode PWM controller with latch off
0 0 protections, LD7532 datasheet. Leadtrend technology document.
 20  20 (2014). [Online]. Available: http://www.leadtrend.com.tw/
 40 Model(Vo=5V)  40 Model(Vo=20V) [7] On-Bright technology. High performance current mode PWM
Experiment(VFPCM) Experiment(VFPCM) controller with peak load, OB2283 datasheet. On-Bright technology
 60  60
1 10 100 110
3
110
4
110
5
1 10 100 110
3
110
4
110
5
document. (2016). [Online]. Available: http://www.on-bright.com/
Frequency(Hz) Frequency(Hz) [8] Texas Instruments. Constant-voltage constant-current flyback
Gvc,spf (s) Gvc,spf (s) controller using opto-coupled feedback, UCC28740 datasheet. Texas
90 90
Instruments document. (2014). [Online]. Available:
45 45
http://www.ti.com/
Phase(degree)

Phase(degree)

0 0 [9] L. Huber and M. M. Jovanović, “Small-signal analysis of DCM flyback


 45  45 converter in frequency-foldback mode of operation,” in Proc. IEEE
 90  90
APEC, 2013, pp. 1746-1752.
[10] B. T. Irving, Y. Panov and M. M. Jovanović, “Variable-frequency
 135 Model(Vo=5V)  135 Model(Vo=20V)
Experiment(VFPCM) Experiment(VFPCM) flyback converter,” in Proc. IEEE APEC, 2003, pp. 977-982.
 180
1 10 100 110
3
110
4
110
5
 180
1 10 100 110
3
110
4
110
5 [11] S. Y. Chen, “Small-signal model for a flyback converter with peak
Frequency(Hz) Frequency(Hz) current mode control,” IET Power Electronics, vol. 7, no. 4, pp. 805-
810, Apr. 2014.
Fig. 27 Model verification by experiment in Vin=265Vrms [12] F. He, “USB Port and power delivery: An overview of USB port
interoperability,” in Proc. IEEE ISPCE, 2015, pp. 1-5.
VI. CONCLUSIONS [13] Y. C. Kang, C. C. Chiu, M. Lin, C. P. Yeh, J. M. Lin, K. H. Chen,
“Quasi-resonant control with a dynamic frequency selector and
A novel, accurate small-signal model of flyback constant current startup technique for 92% peak efficiency and 85%
converter in CCM with VFPCM is proposed. The proposed light-load efficiency fyback converter,” IEEE Trans. Power Electron.,
vol. 29, no. 9, pp. 4959-4969, Sep. 2014.
model resolves the possible unstable problem of the system [14] J. Park, Y. J. Moon, M. G. Jeong, J. G. Kang, S. H. Kim, J. C. Gong,
and provides an easy design of compensation. The describing C. Yoo, “Quasi-resonant (QR) controller with adaptive switching
function approach is used to derive the model for the frequency reduction scheme for flyback converter,” IEEE Trans. Ind.
nonlinear part of the control. Both control behaviors and Electron., vol. 63, no. 6, pp. 3751-3581, June 2016.
characteristics of VFPCM are analyzed and compared with [15] J. Li and F. C. Lee, “New modeling approach and equivalent circuit
representation for current-mode control,” IEEE Trans. Power Electron.,
those of CFPCM to give engineers insight. The modeling vol. 25, no. 5, pp. 1218-1230, May 2010.
results show that there is difference in the DC gain of control [16] S. Tian, F. C. Lee, Q. Li and Y. Yan, “Unified equivalent circuit model
to output transfer function between VFPCM and CFPCM. and optimal design of V2 controlled buck converters,” IEEE Trans.
Therefore, it results in different behaviors of load transient Power Electron., vol. 31, no. 2, pp. 1734-1744, Feb. 2016.
response and frequency response for both types of control. [17] Y. Yan, F. Lee, P. Mattavelli, and P. H. Liu, “Average current mode
control for switching converters,” IEEE Trans. Power Electron., vol.
Steady-state frequency characteristics of VFPCM are also 29, no. 4, pp. 2027–2036, Apr. 2014.
presented. A control IC and a flyback converter are built for [18] S. F. Hsiao, D. Chen, C. J. Chen, and H. S. Nien, “A new multiple-
experimental validation of the proposed model. The model frequency small-signal model for high-bandwidth computer v-core
predictions are in excellent agreement with the simulation regulator applications,” IEEE Trans. Power Electron., vol. 29, no. 4,
and experimental results. pp. 2027–2036, Apr. 2014.
[19] R. D. Middlebrook and S. Cuk, “A general unified approach to
modeling switching-converter power stages,” in Proc. IEEE PESC,
1976, pp. 18-34.

0885-8993 (c) 2016 IEEE. Personal use is permitted, but republication/redistribution requires IEEE permission. See http://www.ieee.org/publications_standards/publications/rights/index.html for more information.
This article has been accepted for publication in a future issue of this journal, but has not been fully edited. Content may change prior to final publication. Citation information: DOI 10.1109/TPEL.2017.2716830, IEEE
Transactions on Power Electronics

[20] R. Erickson and D. Maksimovic, “Fundamentals of Power Electronics,


“ Kluwer Academic Publishers, Norwell, MA, 2001.

0885-8993 (c) 2016 IEEE. Personal use is permitted, but republication/redistribution requires IEEE permission. See http://www.ieee.org/publications_standards/publications/rights/index.html for more information.

You might also like