(Power Systems) Wen-Wei Chen, Jiann-Fuh Chen (Auth.) - Control Techniques For Power Converters With Integrated Circuit-Springer Singapore (2018)

Download as pdf or txt
Download as pdf or txt
You are on page 1of 195

Power Systems

Wen-Wei Chen
Jiann-Fuh Chen

Control
Techniques for
Power Converters
with Integrated
Circuit
Power Systems
More information about this series at http://www.springer.com/series/4622
Wen-Wei Chen Jiann-Fuh Chen

Control Techniques
for Power Converters
with Integrated Circuit

123
Wen-Wei Chen Jiann-Fuh Chen
Department of Electrical and Computer Department of Electrical Engineering
Engineering National Cheng Kung University
Virginia Tech Tainan
Blacksburg Taiwan
USA

ISSN 1612-1287 ISSN 1860-4676 (electronic)


Power Systems
ISBN 978-981-10-7003-7 ISBN 978-981-10-7004-4 (eBook)
https://doi.org/10.1007/978-981-10-7004-4
Library of Congress Control Number: 2017955664

© Springer Nature Singapore Pte Ltd. 2018


This work is subject to copyright. All rights are reserved by the Publisher, whether the whole or part
of the material is concerned, specifically the rights of translation, reprinting, reuse of illustrations,
recitation, broadcasting, reproduction on microfilms or in any other physical way, and transmission
or information storage and retrieval, electronic adaptation, computer software, or by similar or dissimilar
methodology now known or hereafter developed.
The use of general descriptive names, registered names, trademarks, service marks, etc. in this
publication does not imply, even in the absence of a specific statement, that such names are exempt from
the relevant protective laws and regulations and therefore free for general use.
The publisher, the authors and the editors are safe to assume that the advice and information in this
book are believed to be true and accurate at the date of publication. Neither the publisher nor the
authors or the editors give a warranty, express or implied, with respect to the material contained herein or
for any errors or omissions that may have been made. The publisher remains neutral with regard to
jurisdictional claims in published maps and institutional affiliations.

Printed on acid-free paper

This Springer imprint is published by Springer Nature


The registered company is Springer Nature Singapore Pte Ltd.
The registered company address is: 152 Beach Road, #21-01/04 Gateway East, Singapore 189721, Singapore
Contents

1 Introduction . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 1
1.1 Power Converters with Integrated Circuits . . . . . . . . . . . . . . . . . . 1
1.2 Major Control Modes for Power Converters . . . . . . . . . . . . . . . . . 14
1.3 Load Transient Response and Load Regulation . . . . . . . . . . . . . . 17
1.4 Efficiency . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 22
References . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 34
2 Review of the PWM Control Circuits for Power Converters . . . . . . 37
2.1 Voltage-Mode Control Circuit for Power Converters . . . . . . . . . . . 37
2.2 Current-Mode Control Circuit for Power Converters . . . . . . . . . . . 45
2.3 Compensation Design for Power Converters . . . . . . . . . . . . . . . . . 55
2.4 Experimental Results . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 67
2.5 Summary . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 78
References . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 78
3 Designing a Dynamic Ramp with Invariant Inductor
in Current-Mode Control Circuit for Buck Converter . . . . . . . . . . . 81
3.1 Challenges for Wide Input Voltage Range . . . . . . . . . . . . . . . . . . 81
3.2 Dynamic Slope Compensation Design . . . . . . . . . . . . . . . . . . . . . 87
3.3 Experimental Results . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 94
3.4 Summary . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 101
References . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 101
4 Review of the Adaptive On-time Control Circuits for Buck
Converters . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 103
4.1 Adaptive On-time Control Circuits for Buck Converters . . . . . . . . 103
4.2 Ripple-Based Adaptive On-time Control Circuits with Virtual
Inductor Current Ripple for Buck Converters . . . . . . . . . . . . . . . . 108
4.3 Current-Mode Adaptive On-time Control Circuit for Buck
Converters . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 111

v
vi Contents

4.4 Adaptive On-time Control Circuits with Adaptive Voltage


Positioning Design for Voltage Regulators . . . . . . . . . . . . . . . . . . 117
4.5 Summary . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 119
References . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 120
5 Adaptive On-time Control Circuit for Buck Converters . . . . . . . . . . 123
5.1 Increasing Light Load Efficiency with PSM Mode . . . . . . . . . . . . 123
5.2 On-time Generator Circuit of the Adaptive On-time Control
Circuits for Buck Converters . . . . . . . . . . . . . . . . . . . . . . . . . . . . 128
5.3 Comparison of Quick Dynamic Response and Conventional
Quick Response of the On-time Generator Circuit . . . . . . . . . . . . 133
5.4 Experimental Results . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 137
5.5 Summary . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 142
References . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 143
6 Ripple-Based Constant Frequency On-time Control Circuit with
Virtual Inductor Current Ripple for Buck Converters . . . . . . . . . . . 145
6.1 Challenges for Adaptive On-time Control Circuits for Buck
Converters . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 145
6.2 Implemented Control Circuits of the Victual Inductor
Current Ripple . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 148
6.3 On-time Generator Circuit of the Constant Frequency
On-time Control Circuits for Buck Converters . . . . . . . . . . . . . . . 156
6.4 Comparison of Quick Dynamic Response and Conventional
Quick Response of the On-time Generator Circuit . . . . . . . . . . . . 158
6.5 Experimental Results . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 162
6.6 Summary . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 167
References . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 167
7 Constant Current Ripple On-Time Control Circuit
With Native Adaptive Voltage Positioning Design
for Voltage Regulators . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 171
7.1 Challenges for Voltage Regulators . . . . . . . . . . . . . . . . . . . . . . . . 171
7.2 Native Adaptive Voltage Positioning Design for Voltage
Regulators . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 174
7.3 On-Time Generator Circuit of the Constant Current Ripple
On-Time Control Circuit for Voltage Regulators . . . . . . . . . . . . . 177
7.4 Comparison of Quick Dynamic Response and Conventional
Quick Response of the On-Time Generator Circuit . . . . . . . . . . . . 180
7.5 Experimental Results . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 183
7.6 Summary . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 188
References . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 188
Chapter 1
Introduction

1.1 Power Converters with Integrated Circuits

Power converter technology is basic and important because it supports and delivers
all types of electronic equipment and devices, such as consumer electronics,
automotive, and telecommunication [1–11]. Switching and linear power converters
are characterized by small size, light weight, low cost, high reliability, and high
efficiency. Mainstream technologies and trend developments are used in power
converters at present. Owing to the advancement of information and rise of a
communication-oriented society, trends toward personalization and mobilization
have become popular, and the demand for electronic devices that are small, light-
weight, inexpensive, highly reliability, and highly efficient is increasing. Power
converter technology is widely used to meet output loading requirements.
DC–DC power converters are utilized in portable electronic devices, such as
cellular phones and laptop computers, which are primarily supplied with power by
batteries or adapters. Such electronic devices often contain several sub-circuits to
deliver auxiliary power, a central processing unit (CPU) voltage regulator, or a
microprocessor, and the voltage level requirement of each differs from that supplied
by the battery, adapter, or external power supply. In addition, battery voltage
declines as its stored energy is drained. DC–DC power converters increase voltage
from a partially reduced battery voltage and save space compared with using
multiple batteries to accomplish the same task.
Figure 1.1 shows a brief circuit diagram of DC–DC power converters. DC–DC
power converters can achieve power conversion from the input terminal to the
output terminal. The input terminal of DC–DC power converters is the power
source, and DC–DC power converters can regulate output voltage VOUT to a
specified level. Input capacitor CIN is used to filter the input power source and is
placed close to the VIN pin of the integrated circuit (IC) on the evaluation board.
This method can prevent the voltage drop of the printed circuit board (PCB) trace
and ensures that the input voltage possesses good noise immunity. Output capacitor

© Springer Nature Singapore Pte Ltd. 2018 1


W.-W. Chen and J.-F. Chen, Control Techniques for Power
Converters with Integrated Circuit, Power Systems,
https://doi.org/10.1007/978-981-10-7004-4_1
2 1 Introduction

IIN
DC-DC VOUT
Power Converters
VIN
RCIN RCO IOUT

CIN CO

Fig. 1.1 A brief circuit diagram of DC–DC power converters

CO is used to filter the output voltage and provide output capacitor CO energy,
thereby preventing output voltage VOUT from dropping significantly at the load
transient. Output capacitor CO is placed close to the VOUT or VFB pin of IC on the
evaluation board. This method can prevent the voltage drop of the PCB trace from
affecting feedback voltage VFB and ensures that feedback voltage VFB possesses
good noise immunity.
Power ICs are self-contained circuits with many separate components, such as
transistors, diodes, resistors, and capacitors, fabricated into a single tiny chip of a
semiconductor material [12, 13]. ICs are extremely small; they are thousands of
times smaller than discrete circuits. ICs are also lightweight because of the
miniaturized circuits and inexpensive because of the simultaneous production of
hundreds of similar circuits on a small semiconductor wafer. An IC costs as much
as an individual transistor because ICs are mass produced. ICs are highly reliable
because of the absence of soldered joints and the need for only a few intercon-
nections. ICs are widely used in DC–DC power converters because of their
advantages.
DC–DC power converters with ICs are widely utilized in different industries,
such as consumer electronics, automotive, telecommunication, networking, and
medical. The various types of power ICs used at present include voltage regulators,
battery management ICs, integrated application-specific standard product (ASSP)
power ICs, and motor control ICs. Technological advancements and the increasing
demand for battery-operated devices are the major factors that drive the market at
present. The market is expected to grow because the demand for consumer elec-
tronics and automobiles is expected to increase in the future.
DC–DC power converters can be classified as step-down and step-up converters
according to the difference between input and output voltages. A power converter
whose input voltage VIN is larger than output voltage VOUT is called a step-down
converter, whereas a power converter whose output voltage VOUT is larger than
input voltage VIN is called a step-up converter. Several exceptions include
high-efficiency light-emitting diode (LED) driver [14–19], which is a type of DC–
DC power converter that regulates the output load current to drive LEDs. Another
example is charge pumps, which are designed to generate an output voltage that is
double or triple the input voltage.
1.1 Power Converters with Integrated Circuits 3

IIN SMAIN VOUT

IQ
VIN IOUT
RCIN RD1 RCO

VREF
CIN VFB CO

RD2

Fig. 1.2 Circuit diagram of LDO

The two types of DC–DC step-down converters are the switching buck converter
and low-dropout linear regulator (LDO) [20–26]. Figure 1.2 shows a circuit dia-
gram of LDO. LDO possesses a simple circuit structure because it usually consists
of reference voltage VREF, an amplifier, feedback resistors RD1 and RD2, and power
switch SMAIN. LDO regulates DC voltage for the output voltage by reducing the
input voltage across a power switch SMAIN. The working principle of the LDO
regulator involves sampling the output voltage through a resistive divider, which is
fed to the inverting input of the error amplifier. The non-inverting input is tied to a
reference voltage derived from an internal bandgap reference. The error amplifier
always forces the voltages at its input to be equal. The voltage drop across the series
power switch SMAIN is controlled by the error amplifier’s output to control the
output voltage. The output voltage is expressed as Eq. (1.1).
 
RD1
VOUT ¼ 1þ  VREF ð1:1Þ
RD2

The advantage of LDO is that it requires only three parts: power switch SMAIN,
input capacitors CIN, and output capacitors CO. Several LDO IC products include
resistors RD1 and RD2 inside their chip with a fixed output voltage. LDOs are
usually cheaper and much less noisy than inductive switchers. Device input current
IIN is equal to the sum current between output load current IOUT and quiescent
current IQ required by the LDO for its internal circuitry, as shown in Eq. (1.2).
Quiescent current IQ is the current drawn by the LDO to control its internal circuitry
for proper operation. Series power switch SMAIN and ambient temperature are the
primary contributors to quiescent current. If the LDO is operated at a heavy load
and output load current IOUT is larger than quiescent current IQ, quiescent current IQ
will not affect efficiency. However, if the LDO is operated at a light load and output
load current IOUT is smaller than quiescent current IQ, quiescent current IQ will
affect efficiency and must be considered. The efficiency of the solution depends on
the output-to-input voltage ratio in Eqs. (1.3)–(1.5). Given these features, the
4 1 Introduction

advantages of LDO linear regulators over other DC–DC regulators include the
absence of switching noise, reduced IC die size, and design simplicity. Another
advantage of LDO is its low cost. It does not require an inductor, so it saves cost.

IQ ¼ IIN  IOUT ð1:2Þ

PLOSS ¼ ðVIN  VOUT Þ  IOUT þ VIN  IQ ð1:3Þ

PIN ¼ VIN  IIN ð1:4Þ

PIN  PLOSS
g¼ ð1:5Þ
PIN

The main drawback of LDO is its low efficiency for high input-to-output voltage
ratios or high-current applications. Most of the power is dissipated by power switch
SMAIN by Eq. (1.6). Power dissipation PPS depends on the difference between input
and output voltages and output load current IOUT. Therefore, LDO is not an ideal
solution when the difference between input and output voltages is large and when
the output load current is high. Applications with large output load current require
heat sinking, which increases the solution size. For these reasons, LDO is widely
used for applications with low input-to-output voltage ratios and low current, such
as auxiliary power delivery. Auxiliary power delivery is used to obtain a light load
and high noise immunity with a high power supply rejection ratio (PSRR), as
shown in Eq. (1.7). Although LDO obtains a low output voltage ripple without a
switching voltage ripple, a good PSRR must be considered when selecting a suit-
able LDO.
Figure 1.3 shows the perturbation signal injection circuit at the input voltage to
measure PSRR for LDO. AC perturbation signal VAC can be injected by a trans-
former and the bandwidth of the transformer should cover a range between 100 Hz
and 1 MHz to obtain a correct PSRR bode plot. To achieve impedance matching,
the output terminal should be placed at 50 X parallel to the transformer. To obtain a
correct PSRR bode plot, input capacitors CIN should be removed so that the LDO

Perturbation IIN SMAIN VOUT


Injection Circuit
IQ
VAC
IOUT
50 RD1 RCO

VREF
VFB CO
VIN

RD2

Fig. 1.3 Perturbation signal injection circuit at the input voltage to measure PSRR for LDO
1.1 Power Converters with Integrated Circuits 5

can obtain its own PSRR without the capacitor’s influence. Otherwise, AC per-
turbation signal VAC may be filtered by input capacitors CIN. The two output
terminals of a network analyzer or spectrum analyzer should be connected to ter-
minals VIN and VOUT to obtain a PSRR bode plot individually.

PPS ¼ ðVIN  VOUT Þ  IOUT ð1:6Þ


 
VIN Ripple

PSRR ¼ 20  log  ð1:7Þ
VOUT Ripple

Most LDO IC products have a relatively high PSRR at a low frequency range of
100 Hz–1 kHz. Having a high PSRR over a wide band allows LDO to reject
high-frequency noise, such as that arising from power switch SMAIN. Similar to
other specifications, PSRR fluctuates over frequency, temperature, current, and
output voltage. Figure 1.4 shows the experimental results of PSRR for LDO at
different output load currents (1 and 100 mA). The red-colored curve represents the
measurement results for the PSRR bode plot at an output load current of 1 mA, and
the blue-colored curve represents the measurement results for the PSRR bode plot
at an output load current of 100 mA. The operation conditions are as follows: 3.3 V
of input voltage VIN, 2.5 V of output voltage VOUT, 10 mV of AC perturbation
signal VAC, and 1 and 100 mA of output load current. At the output load current of
1 mA, LDO can obtain a PSRR of 60 dB, which is similar to that at the output load
current of 100 mA.
At the frequency range of 100 Hz–450 kHz, the range of the PSRR results
extends from bandgap filter roll-off frequency to unity-gain frequency, where PSRR
is dominated by the open-loop gain of LDO. Within the frequency range of
450 kHz–1 MHz, the range of the PSRR results exceeds the unity-gain frequency,
where feedback loop exerts a negligible effect; therefore, the output capacitor

-10

-20
PSRR (dB)

-30

-40

-50
Output Load Current=1mA
-60
Output Load Current=100mA
-70
100 1000 10000 100000 1000000
Frequency (Hz)

Fig. 1.4 Experimental results of PSRR for LDO at different output load currents (1 and 100 mA)
6 1 Introduction

IIN L IL VOUT
SA VLX

VIN IOUT
RCIN RD1 RCO

DB
CIN VUG VFB CO

RD2

Fig. 1.5 Circuit diagram of an asynchronous buck converter

dominates along with the parasitic impedance from the input and output terminals.
The gate driver’s capability to drive power switch SMAIN at the frequency range of
450 kHz–1 MHz is also an influencing factor.
The switching buck converter is the other type of DC–DC step-down converter.
Figure 1.5 shows a circuit diagram of an asynchronous buck converter. A buck
converter is applied to reduce DC voltage from the input voltage to the output
voltage by using only power switches, main inductor L, input capacitors CIN, and
output capacitors CO. Similar to the setup in any switching power converter, a
power switch controls the transfer of energy. Inductors and capacitors are also used
to store energy. Power switches are realized by using power semiconductor devices,
such as metal–oxide–semiconductor field-effect transistor (MOSFET) SA and diode
DB, which are controlled to turn on and turn off as required to regulate output
voltage VOUT. Usually, a P-channel MOSFET (PMOS) is used as MOSFET SA
instead of NMOS because if NMOS is employed as a high-side switch, driving it
would be difficult because the gate and source are connected to the voltage supply.
However, PMOS needs more internal circuits than NMOS. In the selection of diode
DB, three key specifications must be checked: reverse voltage, forward voltage
drop, and forward current. The reverse voltage must be higher than the maximum
voltage of voltage terminal VLX to avoid damaging the diode. The forward voltage
drop should be small. A large forward voltage drop can result in a large conduction
loss. Meanwhile, the forward current must be larger than the maximum output load
current; otherwise, it cannot provide sufficient energy to the output loading
requirements. In addition, the inductor current ripple should be considered in the
forward current rating for diode DB. The forward current should be larger than the
sum current between a half of inductor current ripple and the maximum output load
current.
To achieve high load efficiency, diode DB is usually replaced by n-channel
MOSFET SB as a low-side switch to increase load efficiency because NMOS can
save more IC die size than PMOS. The total power loss in a DC–DC converter is
significantly reduced by this replacement because the voltage drop in conducted
1.1 Power Converters with Integrated Circuits 7

IIN L IL VOUT
SA VLX

VIN IOUT
RCIN RD1 RCO
SB
CIN VUG VLG VFB CO

RD2

Fig. 1.6 Circuit diagram of a synchronous buck converter

NMOS is very low in comparison with that in a conducted diode (even the Schottky
diode, which has a low forward voltage drop). A circuit diagram of a synchronous
buck converter is shown in Fig. 1.6. The main advantage of a low-side switch is
that the voltage drop across the low-side MOSFET can be lower than the voltage
drop across the power diode of the asynchronous converter. When no change in the
current level occurs, a low voltage drop translates to low power dissipation and high
efficiency. For a synchronous buck converter, dead-time control must be imple-
mented to prevent shoot-through current from flowing through the main power
MOSFET during switching transitions by controlling the duty cycle of the
MOSFET drivers. The high-side driver is not allowed to turn on until the gate drive
voltage to the low-side MOSFET is low, and the low-side driver is not allowed to
turn on until the voltage at the junction of the power MOSFET is low.
Figure 1.7 shows the control signals for buck converter operation. When
high-side switch SA is turned on by control signal VUG, a path is provided for the
DC input voltage to charge the inductor and supply the output loading require-
ments. Charging continues until feedback voltage VFB reaches reference voltage
VREF. Then, the control part turns off the high-side switch to keep feedback voltage
VFB close to internal reference voltage VREF. Therefore, no path exists to charge the
inductor. Afterward, the inductor changes its voltage polarity, and the current flows
in the same direction through low-side switch SB, which is turned on by the control
signal VLG part with dead-time control. Discharging continues until feedback
voltage VFB is less than internal reference voltage VREF. The control part again turns
on high-side switch SA to compensate for the output voltage drop. This cycle
continues until complete regulation of output voltage is achieved.
This process is accomplished by sensing the output voltage of the circuit by
means of a negative feedback loop that adjusts the duty cycle by Eq. (1.8) to control
the on and off states of the MOSFET switches via control signals VUG and VLG
under specified switching frequency FS. In Eq. (1.8), TON is the time interval that
power MOSFET SA conducts (on state), TOFF is the time interval that power
MOSFET SB conducts (off state), TS is the period, and FS is the switching
8 1 Introduction

Fig. 1.7 Control signals for V


buck converter operation
VUG

0
t
VLG
Dead-Time

0
t
VLX VIN

0
t
VOUT

VOUT
t
VFB
VFB
t
A

IL ΔIL
IOUT

frequency. Switching frequency FS of control signals VUG and VLG generally lies in
the range of 100 kHz–2 MHz.
VOUT TON TON
D¼ ¼ ¼ ¼ FS  TON ð1:8Þ
VIN TS TON þ TOFF

The output voltage is also designed similar to that in an LDO circuit, as shown in
Eq. (1.1). The inductor value and operating frequency determine inductor current
ripple DIL according to specific input and output voltages. Inductor current ripple
DIL has a positive correlation with input voltage. Thus, if the input voltage is
increased, inductor current ripple DIL can also be increased by Eq. (1.9). Having a
low DIL reduces not only ESR losses in the output capacitors but also the output
voltage ripple. High frequency with a small ripple current can result in high-
efficiency operations. In general, inductor current ripple DIL is equal to 20% of the
1.1 Power Converters with Integrated Circuits 9

maximum load current. In poor conditions, the maximum input voltage can obtain
the largest inductor current ripple DIL, so the inductor value can be determined by
Eq. (1.10).
 
VOUT VOUT
DIL ¼  1 ð1:9Þ
FS  L VIN
 
VOUT VOUT
L¼  1 ð1:10Þ
FS  DIL MAX VIN MAX

Moreover, the selection of output capacitor CO is determined by the required


ESR RCO to minimize the voltage ripple. The amount of bulk capacitance is also a
key factor in CO selection because it ensures that the control loop is stable. Output
voltage ripple DVOUT is determined by Eq. (1.11). DVOUT also depends on inductor
current ripple DIL. Therefore, output voltage ripple DVOUT has a positive correlation
with input voltage because inductor current ripple DIL has a positive correlation
with input voltage. Increasing the input voltage would increase output voltage
ripple DVOUT. Multiple capacitors placed in parallel may need to meet ESR and
RMS current handling requirements. In addition, ceramic capacitors are widely
used as output capacitors because they are inexpensive and highly reliable and
possess low ESR and small size. Equation (1.11) can be modified and rewritten as
Eq. (1.12) without affecting ESR.
 
1
DVOUT  DIL  RCO þ ð1:11Þ
8  FS  CO

1
DVOUT  DIL  ð1:12Þ
8  FS  CO

For many power IC products and applications, the evaluation board size can be
reduced by selecting a device in which the IC switching MOSFET elements are
designed and embedded by a real implemented internal circuit. This method is
needed to check the thermal of IC die temperature.
The output voltage of a buck converter must be smaller than the input voltage, so
the output voltage is still smaller than the input voltage during the soft start time of
a buck converter. A buck converter does not need the pre-charge function to reg-
ulate the output voltage. The soft start time is designed to reduce the inrush current
at the input terminal. Figure 1.8 shows the experimental results of soft start with
input voltage for buck converter. Input current IIN is measured with an oscilloscope
and is very small with a soft start time of 1.3 ms. In general, the current limit at the
low-side MOSFET SB function can limit the input energy to the output terminal.
Hence, input current IIN should be reduced at the soft start time. In addition, internal
reference voltage VREF is controlled with a slow slew rate. These two means can
reduce the inrush current at the input terminal.
10 1 Introduction

VLX

VIN

VOUT
Soft Start
IIN

Fig. 1.8 Experimental results of soft start with input voltage for buck converter

Figure 1.9 shows a circuit diagram of the synchronous boost converter. A boost
converter is applied to increase the DC voltage from the input voltage to the output
voltage by using only power switches, main inductor L, input capacitors CIN, and
output capacitors CO. As in any switching power converter, a power switch controls
the transfer of energy. Inductors and capacitors are also used to store energy. Power
switches are realized by using power semiconductor devices, such as MOSFET SA
and MOSFET SB, which are controlled to turn on and turn off as required to
regulate output voltage VOUT. Usually, PMOS is preferred to be used as MOSFET
SA instead of NMOS, similar to buck converter. The n-channel MOSFET SB as a
low-side switch increases the load efficiency. The main advantage of a low-side
switch is that the voltage drop across the low-side MOSFET can be lower than the

IIN L IL VOUT
VLX SA

VIN IOUT
RCIN RD1 RCO
SB VUG
CIN VLG VFB CO

RD2

Fig. 1.9 Circuit diagram of the synchronous boost converter


1.1 Power Converters with Integrated Circuits 11

Fig. 1.10 Control signals in V


the boost converter operation
VUG

0
t
VLG
Dead-Time

0
t
VLX VIN

0
t
VOUT

VOUT
t
VFB
VFB
t
A

IL ΔIL

voltage drop across the power diode of the asynchronous converter. When no
change in the current level occurs, a low voltage drop translates to low power
dissipation and high efficiency.
Figure 1.10 shows the control signals in the boost converter operation. When
low-side switch SB is turned on by control signal VLG, a path is provided for the DC
input voltage to charge the inductor and cut the supply of the output loading
requirement. Charging continues until feedback voltage VFB reaches reference
voltage VREF. Then, the control part turns off the low-side switch to keep feedback
voltage VFB close to internal reference voltage VREF. Therefore, no path exists to
charge the inductor. The inductor then changes its voltage polarity, and the current
flows in the same direction through high-side switch SA, which is turned on by the
control signal VUG part with dead-time control. Discharging continues until
12 1 Introduction

feedback voltage VFB is less than internal reference voltage VREF. The control part
again turns on low-side switch SB to compensate for the output voltage drop. This
cycle continues until complete regulation of output voltage is achieved.
This process is accomplished by sensing the output voltage of the circuit by
means of a negative feedback loop that adjusts the duty cycle by Eq. (1.13) to
control the on and off states of the MOSFET switches via control signals VUG and
VLG under specified switching frequency FS. In Eq. (1.13), TON is the time interval
that power MOSFET SB conducts (on state), TOFF is the time interval that power
MOSFET SA conducts (off state), TS is the period, and FS is the switching
frequency.

VOUT VIN TON TON


D¼ ¼ ¼ ¼ FS  TON ð1:13Þ
VOUT TS TON þ TOFF

The output voltage of the boost converter is also designed similar to that of buck
converter by using Eq. (1.1). The inductance depends on the maximum input
current because the inductor current is equal to the input current, as shown in
Eq. (1.14). As a general rule, the inductor current ripple is designed around 20% to
40% of the maximum input current. Assuming that the inductor current ripple is
equal to 40% of the input current, it can be obtained with Eq. (1.15). To select a
suitable inductor, the inductor saturated current should be larger than the sum
current between a half of inductor current ripple and the maximum inductor current.
The inductance should be determined, as shown in Eq. (1.16), where FS is the
switching frequency. To consider system performance, a shielded inductor is pre-
ferred to avoid the EMI issue.

VOUT  IOUT MAX


IL MAX ¼ ð1:14Þ
g  VIN

DIL RIPPLE ¼ 40%  IL MAX ð1:15Þ

g  VIN2  ðVOUT  VIN Þ


L¼ ð1:16Þ
2 I
40%  VOUT OUT MAX  FS

The selection of output capacitor CO is determined by the required ESR RCO to


minimize the voltage ripple. The amount of bulk capacitance is also a key factor in
CO selection to ensure that the control loop is stable. Output voltage ripple DVOUT
is determined by Eq. (1.17). DVOUT also depends on output load current IOUT and
inductor current IL. Multiple capacitors placed in parallel may need to meet ESR
and root-mean-square (RMS) current handling requirements. Ceramic capacitors
are widely used as output capacitors because they are inexpensive and highly
reliable and possess low ESR and small size. Equation (1.17) can be modified and
rewritten as Eq. (1.18) without affecting ESR.
1.1 Power Converters with Integrated Circuits 13

D  IOUT
DVOUT ¼ DIL  RCO þ ð1:17Þ
g  FS  CO

D  IOUT
DVOUT ¼ ð1:18Þ
g  FS  CO

The output voltage of the boost converter must be larger than the input voltage,
so the output voltage may be larger than the input voltage during the soft start time
of the boost converter. However, the output voltage is zero in the initial condition,
so most IC products for the boost converter have a pre-charge function. This
function can achieve an output voltage that is equal to the input voltage. The output
voltage can then be regulated to meet the user design rating. The boost converter
not only has a soft start time similar to buck converter; it also has a pre-charge time.
Soft start time and pre-charge time are designed to reduce the inrush current at
the input terminal. Figure 1.11 shows the experimental results of soft start with
input voltage for the boost converter. The output voltage is equal to 3.3 V, similar
to the input voltage at pre-charge time. The pre-charge time is close to 2 ms. Based
on the pre-charge implemented circuit, three methods can be adopted to achieve an
output voltage that is equal to the input voltage with a low inrush current at the
input terminal. These methods should be avoided for the high-side MOSFET SA to
fully turn on. The first method senses the high-side MOSFET SA to clamp its
current and reduce inductor current IL. The second method produces the other
couple control signals of VUG and VLG to deliver input energy to the output terminal
slowly. The last method is similar to the second method and reduces the switching
frequency for the other couple control signals of VUG and VLG to deliver input
energy to the output terminal slowly.
Inductor current IL is equal to input current IIN and is measured with an oscil-
loscope. Inductor current IL is very small with a soft start time of 1.3 ms. In general,
the current limit at the low-side MOSFET SB function can limit the input energy to

Fig. 1.11 Experimental


results of soft start with input
voltage for boost converter VIN

VOUT
Pre-charge Soft Start
IL
14 1 Introduction

the output terminal. Hence, inductor current IL should be reduced at soft start time.
In addition, internal reference voltage VREF can be controlled with a slow slew rate.
These two means can reduce the inrush current at the input terminal.

1.2 Major Control Modes for Power Converters

The major control modes according to the couple control signals of VUG and VLG
can be classified as pulse width modulation (PWM) control mode and pulse fre-
quency modulation (PFM) control mode [1–11]. The PWM control mode can
generate couple control signals of VUG and VLG for a fixed switching frequency.
The switching frequency is a constant cycle by cycle at the steady status.
Figure 1.12 shows the PWM control mode to generate couple control signals for
VUG and VLG. The PWM control mode is widely used to control the duty cycle
regulating the output voltage to meet the user design. For power IC products, the
current-mode and voltage-mode control circuits belong to the PWM control mode
because both circuits use a clock signal VCLOCK to make a fixed switching fre-
quency. The pulse of VCLOCK is used to trigger the SR flip-flop, and the input
terminal begins to deliver energy to the output terminal until the energy is sufficient
at the output terminal and feedback voltage VFB reaches reference voltage VREF. The
control circuit then generates another pulse to reset the SR flip-flop.
The PFM control mode can generate the couple control signals of VUG and VLG
for a variable switching frequency, so the on-time width or off-time width is the
same width cycle by cycle at the steady status. When the output terminal needs

Fig. 1.12 PWM control V


mode to generate the couple
control signals for VUG and
VUG
VLG

0
Fixed FS Fixed FS Fixed FS
t
VLG
Dead-Time

0
t
VCLOCK

0
t
1.2 Major Control Modes for Power Converters 15

more energy from the input terminal, the PFM control mode may generate one
control signal VTRIG to trigger the on-time generation circuit or off-time generation
circuit. For power IC products, the on-time generation circuit or off-time generation
circuit is designed and implemented by Eq. (1.19).

C  VREFA
TON ¼ ð1:19Þ
I

Figure 1.13 shows a circuit diagram of the on-time generation circuit with SR
flip-flop for PFM control mode. The charged current ICHARGE is calculated by
Eq. (1.20). ICHARGE can charge capacitor C1 when control signal VLG turns off.
Then, voltage VCHARGE is increased linearly. When VCHARGE is larger than refer-
ence voltage VREF2 and the output signal VCMP of the comparator is changed from
low level to high level, the control signals of VUG and VLG can be used to drive and
control power MOSFETs SA and SB.

C  VREF2
TON ¼ ð1:20Þ
ðVREF1
R1 Þ  G1

Figure 1.14 shows a PFM control mode to generate the couple control signals
for VUG and VLG. The PFM control mode is widely used to control the variable
switching frequency regulating the output voltage to meet the user design. A major
difference of PFM control is not having a clock signal VCLOCK. Given that PFM is
not a constant frequency, it does not need VCLOCK to keep the same switching
frequency. If the output terminal needs more energy to keep the output voltage and
feedback voltage VFB is less than internal reference voltage VREF, control signal
VTRIG should be triggered from low level to high level. The pulse of clock signal
VTRIG is used to trigger SR flip-flop, and the input terminal begins to deliver energy
to the output terminal until the capacitor’s voltage VCHARGE is larger than internal
reference voltage VREF2. The on-time generation circuit then controls the output
terminal of comparator VCMP from low level to high level to reset SR flip-flop.

VDD

VREF1 R1
G1
VCHARGE
VCMP
ICHARGE R Q VUG

VLG S1 C1 VREF2
VTRIG S Q' VLG

Fig. 1.13 Circuit diagram of on-time generation circuit with SR flip-flop for PFM control mode
16 1 Introduction

Fig. 1.14 PFM control mode V


to generate the couple control
signals for VUG and VLG
VUG

0
Fixed TON Fixed TON Fixed TON
t
VLG
Dead-Time

0
t
VREF2
VCHARGE

0
t
VCMP

0
VTRIG
t

For power IC products, the conventional constant on-time control circuit and
adaptive on-time control circuit belong to the PFM control mode. The conventional
constant on-time control circuit can generate the same on-time width regardless of
the input or output voltage variation. The switching frequency should be increased
if the output terminal is under a heavy load and the on-time generation circuit
generates more control pulses to deliver more energy from the input terminal to the
output terminal. If the output terminal is under high output voltage and the on-time
generation circuit generates more control pulses to deliver more energy from the
input terminal to the output terminal, the switching frequency should also be
increased. In addition, the switching frequency of conventional constant on-time
control circuit depends on the output load current and output voltage.
Meanwhile, the on-time generation circuit of adaptive on-time control circuit is
used to sample the input and output voltages. The adaptive on-time control circuit
can be classified as constant frequency on-time and constant current ripple on-time
according to the connection form. Constant frequency on-time control circuit is
different from conventional constant on-time (COT) control circuit because it
prevents the generation of the same on-time width at a high output voltage or under
heavy load. Constant current ripple on-time control circuit is also different from
1.2 Major Control Modes for Power Converters 17

conventional COT control circuit because it prevents the generation of the same
on-time width at the voltage drop between the input and output voltages. Therefore,
even when the voltage drop between the input and output voltages is changed, the
inductor current ripple maintains a constant value.

1.3 Load Transient Response and Load Regulation

Switching power supply with power IC is commonly used for fast load transient
response because it can improve the load transient response to reduce output
capacitance, especially in CPU and load applications with high current slew rate.
Most modern consumer electronic products contain embedded CPU functions,
wireless connectivity, and media players. When these electronic products are used
and operated, the output loading requirements are almost like a pulsating load.
A pulsating load generally involves a fast slew rate of rising and falling time. When
consumer electronic products need more energy and the output load current is
changed from light to heavy, it is called load droop. If the energy is sufficient to
supply consumer electronic products and the output load current is also changed
from heavy to light, it is called load release. Load release and load droop belong to
load transient.
For power IC products, load transient is the most important measured item. The
experimental results of load transient are known as IC performance and charac-
teristic. The load transient response is a quick way to determine the power converter
control loop response. The power converter control loop can determine the power
converter regulation speed and reveal control loop stability problems. The load
transient response can address the steady-status error further to determine the load
regulation performance.
Figure 1.15 shows the experimental results for the load transient without a
steady-status error. The operation conditions are as follows: input voltage of 12 V,
output voltage of 1.05 V, switching frequency of 650 kHz, the output capacitor is
22 lF, and the inductor is 1.4 lH. The red-colored waveform represents the output
voltage, and the pink-colored waveform represents the output load current.
According to the experimental results, the output voltage suffers a voltage drop at
the load droop. At the same time, the power IC needs to convert the input power
source to deliver more energy to the output loading requirements to minimize the
output voltage drop. Once the input power source delivers sufficient energy to the
output loading requirements at load release, the output voltage suffers an overshoot.
The extra pulse at load release must be avoided because a large overshoot results in
damage to consumer electronic products.
The experimental results of load transient show the same voltage level of output
voltage under heavy and light loads, so this power converter has no steady-status
error and demonstrates good load regulation performance.
Figure 1.16 shows the experimental results for load regulation without a
steady-status error at an output voltage of 1.05 V under different input voltages of 5,
18 1 Introduction

VOUT

IOUT

Fig. 1.15 The experimental results for the load transient without a steady-status error

Output Voltage vs. Output Load Current


1.048

1.046
Output Voltage (V)

1.044

1.042
Input Voltage: 5V
1.040 Input Voltage: 12V
Input Voltage: 17V
1.038
0 0.5 1 1.5 2 2.5 3
Output Load Current (A)

Fig. 1.16 The experimental results for the load regulation without a steady-status error at an
output voltage of 1.05 V under different input voltages of 5, 12, and 17 V

12, and 17 V. The specification of load regulation is generally defined as +1%/−1%


of the output voltage. Based on this operation condition, the output voltage should
range from 1.0395 V to 1.0605 V. The experimental results meet this specification
of load regulation and can prove that the output load current does not affect the
output voltage regulation. The experimental results are also measured under dif-
ferent input voltages of 5, 12, and 17 V. The experimental results prove that the
input voltage does not affect the output voltage regulation. For these reasons, this
power converter with power IC is suitable for applications with a fixed output
voltage under different output load currents and input voltages.
1.3 Load Transient Response and Load Regulation 19

If the system control loop response is fast, the output voltage may suffer from a
low voltage drop. If the system control loop response is too fast, then the power
converter may suffer from noise jitter, subharmonic, or even result in an unstable
system. If the system control loop response is slow, the output voltage may suffer
from a large voltage drop. A large voltage drop causes consumer electronic prod-
ucts to suffer from damage or failure; hence, the design of the system control loop
response is a trade-off. The system control loop response is an optimal design based
on its operation conditions and must cover the worst operation conditions.
Most IC products have no steady-status error design because the electronic
device or equipment cannot accept a large variation of the output voltage. A large
steady-status error cannot meet the specification of load regulation. CPU requires a
unique power delivery and allows the supply voltage to have a large steady-status
error. CPU generally uses adaptive voltage position (AVP) to achieve the
steady-status error. Output voltage has an inverse correlation with output load
current for CPU with AVP [27–33]. If the output load current is changed from light
to heavy, the output voltage for delivery to the CPU will be reduced. Output
impedance may behave as an equivalent resistant RLL by Eq. (1.21), as shown in
Fig. 1.17. Figure 1.17 shows the voltage identification (VID) VVID voltage by a real
IC implemented circuit, and the user should base on the VID code table to deter-
mine the VVID voltage by the I2C interface. If the CPU’s load current is zero, output
voltage VOUT is close to VVID voltage. The resistor RLL is generally designed with
external components, and the VVID voltage is equal to the sum voltage between
voltage droop VDROOP and output voltage VOUT. The load transient response occurs
in the CPU and should not require much energy to keep and regulate the same
output voltage; hence, the steady-status error can save on energy and capacitors.

VOUT ¼ VVID IOUT  RLL ð1:21Þ

Figure 1.18 shows the experimental results for load transient with a steady-status
error using AVP. The operation conditions are as follows: input voltage of 12 V,
output voltage of 1.1 V, switching frequency of 350 kHz, the output capacitors
consist of three OS-CON capacitors (560 lF) and 22 ceramic capacitors (22 lF),
and the inductor is 360 nH for each phase. The blue-colored waveform represents
the output voltage, and the black-colored waveform represents the output load
current. According to the experimental results, the output voltage suffers from a
voltage drop at load droop. At the same time, the power IC needs to convert the

Fig. 1.17 Equivalent circuit IC Internal


for VVID voltage by a real IC RLL VOUT
implemented circuit with
AVP VDROOP IOUT
VVID
20 1 Introduction

VOUT

35mV

70A

IOUT
35A

Fig. 1.18 The experimental results for the load transient with a steady-status error using AVP

input power source to deliver more energy to the output loading requirements to
minimize the output voltage drop. The output voltage can be controlled and reg-
ulated at a different voltage level according to the output load current. The output
load current is changed from 35 to 70 A, and the output voltage may be changed
from 1.1 to 1.065 V so as not to regulate the same output voltage of 1.1 V. Based
on this experimental result, resistor RLL is defined as 1 mX, and voltage droop
VDROOP is equal to 35 mV.
The experimental results of the load transient show a different voltage level of
the output voltage at heavy and light loads, so this power converter has a
steady-status error using AVP and shows poor load regulation performance.
Figure 1.19 shows the experimental results for load regulation with a
steady-status error at an output voltage of 1.1 V and input voltage of 12 V. The

1.14
1.12
1.10
1.08
VOUT [V]

1.06
1.04
1.02
VOUT
1.00
Lower TOB (Min_SPEC)
0.98 Upper TOB (Max_SPEC)
0.96
0 20 40 60 80 100
IOUT [A]

Fig. 1.19 The experimental results for load regulation with a steady-status error at an output
voltage of 1.1 V and input voltage of 12 V
1.3 Load Transient Response and Load Regulation 21

specification of load regulation generally defines the upper tolerance of the band
(TOB) as the maximum voltage and lower TOB as the minimum voltage of the
output voltage. Based on this operation condition, the output voltage should be
within upper and lower TOB limits. The experimental results meet this specification
of load regulation and can prove that the output load current directly affects the
output voltage regulation. For these reasons, this power converter with power IC is
suitable for applications with a steady-status error using AVP, such as CPU load
delivery.
Figure 1.20 shows a comparison of PWM and PFM control modes at load droop
under the same operation condition [34–36]. Comparing the load transient between
PWM and PFM control modes is difficult because two similar topologies for PWM
and PFM control modes must be determined. Interestingly, the current-mode
control circuit belongs to the PWM control mode, which is very similar to the
current-mode adaptive on-time control circuit. The current-mode adaptive on-time
control circuit belongs to the PFM control mode. The on-time generator of the
current-mode adaptive on-time control circuit is significantly different from that of
the current-mode control circuit. The current-mode adaptive on-time control circuit
is highly similar to the current-mode control circuit. Given the different on-time
generator, it is easy to determine the load transient response between PWM and
PFM control modes at load droop. These simulation results use the same operation

IOUT
(A)

VUG
(V)

VUG
(V)

VOUT
(V)

t (us)

Fig. 1.20 Comparison of PWM and PFM control modes at load droop under the same operation
condition
22 1 Introduction

conditions to compare PWM and PFM control modes at an output voltage VOUT of
1.05 V, input voltage VIN of 19 V, switching frequency FS of 320 kHz, the output
capacitors CO of two PSCAP is 330 lF, the inductor L is 2.2 lH, and light load of 1
A to heavy load of 9 A within 8 ls. The PFM control mode can generate multiple
pulses with a minimum off-time mechanism to prevent VOUT from decreasing
significantly. The PFM control mode is useful for the reduction of VOUT
peak-to-peak voltage at load droop. The PWM control mode can simply increase its
PWM pulse width with the same switching frequency at the droop. Comparison of
PFM and PWM control modes at the droop shows that the former can generate
more PWM pulses than the latter. Therefore, the PFM control mode can achieve a
faster transient response than the PWM control mode. In addition, the PFM control
mode can achieve shorter settling time than the PWM control mode.

1.4 Efficiency

The efficiency of DC–DC power converters with ICs is a critical criterion in dif-
ferent industries, such as consumer electronics, automotive, telecommunication,
networking, and medicine. The efficiency of the selected power solutions relates to
system power loss and the thermal performance of ICs, PCBs, and other compo-
nents, which determines the power effectiveness. High-efficiency DC–DC power
converters with ICs result in minimal heat dissipation, which reduces system cost
and the size of elements, such as heat sinks, fans, and their assembly. For example,
in a battery-operated system, less power loss means that the device can use the same
battery for a longer run time because the device pulls less current from the battery.
An ideal DC–DC power converter has no loss on the components and power
switch; it has no switching and conduction losses. An ideal power switch implies
zero losses, thus offering 100% efficiency. However, the components are not ideal,
as illustrated in the following examples.
In general, to consider the various factors that contribute to efficiency, the focus
should be on buck converter, which is the most popular DC–DC power converter
applied in consumer electronics, automotive, telecommunication, networking, and
medical applications. A high-efficiency buck converter should achieve low power
dissipation. Figure 1.21 shows the power dissipation in the inductor and MosFETs

IIN ISA L IL DCR


SA VLX VOUT

ISB
VIN IOUT
RCIN RCO
SB
CIN VUG VLG CO

Fig. 1.21 The power dissipation in the inductor and MosFETs for buck converter
1.4 Efficiency 23

Fig. 1.22 The inductor V


current IL for buck converter
VLX VIN

0
t
VOUT
VOUT
t
A
IL
ΔIL
IOUT

for buck converter. The three main causes of power dissipation in a buck converter
are inductor conduction loss, MOSFET conduction loss, and MOSFET switching
loss, which occur in the components when buck converter operates in continuous
conduction mode (CCM), fixed switching frequency, fixed input voltage, and fixed
output voltage [37–46].
Figure 1.22 shows the inductor current for buck converter. DCR is the DC
resistor of the inductor. The power dissipation in DCR addresses the inductor
conduction loss in Eq. (1.22). The RMS inductor current is shown in Eq. (1.23).
The RMS inductor current can be written to approximate the output load current as
Eq. (1.24) because inductor current ripple DIL is not a major influencing factor of
RMS inductor current. The inductor selection and output load current are usually
decided by the customers or users, so improving inductor conduction loss is
difficult.

PDCR ¼ IL2 RMS  DCR ð1:22Þ

DIL2
IL2 RMS ¼ IOUT
2
þ ð1:23Þ
12

IL2 RMS IOUT ð1:24Þ


2

MOSFET conduction loss means power switch SA is turned on and experiences


power dissipation. Power switch SB is turned on and also experiences power dis-
sipation. When any power switch is turned on, this power switch can be modeled as
resistor RON. The conduction loss of power switch SA can be calculated by
Eqs. (1.25)–(1.26). The conduction loss of power switch SA occurs when power
24 1 Introduction

Fig. 1.23 The current ISA of V


power switch SA for buck
converter VLX VIN

0
t
VOUT
VOUT
t

A
ISA
ΔIL
IOUT

t
D*TS
TS

switch SA is turned on during D*TS, so current ISA goes through resistor RSA_ON as
shown in Fig. 1.23.

PSA CON ¼ ISA


2
RMS  RSA ON ð1:25Þ
 
VOUT DIL2
PSA CON ¼  IOUT þ
2
 RSA ON ð1:26Þ
VIN 12

The conduction loss of power switch SB can be calculated by Eqs. (1.27)–(1.28).


The conduction loss of power switch SB occurs when power switch SB is turned on
during D′*TS, so current ISB goes through resistor RSB_ON as shown in Fig. 1.24.

PSB CON ¼ I2SB RMS  RSB ON ð1:27Þ


   
VOUT DIL2
PSB CON ¼ 1  IOUT þ
2
 RSB ON ð1:28Þ
VIN 12

The overall conduction loss of MOSFETs is equal to the sum of power dissi-
pation PSA_CON and power dissipation PSB_CON. The conduction loss of MOSFETs
depends on the output load current. For this reason, if the output load current is
heavy load, the conduction loss of MOSFETs should be increased. The efficiency
under heavy load will improve if the conduction loss of MOSFETs is reduced. One
means to reduce the conduction loss of MOSFETs is to reduce their turned-on
resistor RON. However, low resistor RON needs more IC die size, which depends on
1.4 Efficiency 25

Fig. 1.24 The current ISB of V


power switch SB for buck
convert VLX VIN

0
t
VOUT
VOUT
t
A

ISB
ΔIL
IOUT

t
D' *TS
TS

the IC cost. Meanwhile, a large conduction loss results in a large turned-on resistor
RON, which may increase the IC junction temperature. The IC must be checked to
prevent it from suffering from a thermal shutdown.
MOSFETs have a very short switching time, so the switching loss of MOSFETs
originates from dynamic drain-to-source voltage VDS and drain current ID that the
MOSFETs must handle during the time it takes to turn on or off.
Figure 1.25 shows the level shift driver circuit that drives power switch SA. This
level shift driver circuit is applied to change the power domain from IC internal
power to external input voltage VIN. This function of the level shift driver circuit
can provide strong power to control power switch SA. Figure 1.25 shows the
equivalent circuit of power switch SA with internal parasitic capacitors, such as
CGD, CGS, and CDS.
When power switch SA is turned on, internal parasitic capacitor CGS must charge
until the voltage of capacitor CGS is equal to threshold voltage VTH. When the
voltage of capacitor CGS is larger than threshold voltage VTH, current IDS of power
switch SA is increased. Current IDS increases to equal output load current IOUT until
the voltage of capacitor CGS equals the Miller plateau voltage VPLAT within T1 time.
At this moment, the power delivery VIN of the level shift driver circuit charges the
Miller parasitic capacitor CGD so that the voltage of capacitor CGS can be main-
tained for a while. At the same time, voltage VLX is increased to equal input voltage
VIN, and voltage VDS is reduced to zero within T2 time (Fig. 1.26). In Fig. 1.26, the
turn on switching loss PSA_SWON of power switch SA is equal to the power dissi-
pation in the area during T1 and T2 time, as shown in Eqs. (1.29)–(1.32).
26 1 Introduction

VIN

Drain
VIN SA
CGD

SON
Turn-ON
RGD RG
Gate CDS
Duty Cycle SOFF Turn-OFF

CGS

Source

VLX

Fig. 1.25 The level shift driver circuit that drives power switch SA

Fig. 1.26 Turn on switching A


loss PSA_SWON of power
switch SA in the area during IDS IOUT
T1 time and T2 time

t
T1 T2
V

VDS VIN

0
t
VGS
VIN
VPLAT
VTH

QTH QSW_ON t
QGS QGD

T1 þ T2 TSA SWON
PSA SWON ¼ FS  VDS  IDS  ¼ FS  VIN  IOUT  ð1:29Þ
2 2
QSW ON
TSA SWON ¼ ð1:30Þ
ISA SWON
1.4 Efficiency 27

VIN VPLAT
ISA SWON ¼ ð1:31Þ
RSON ON þ RGD þ RG

QSW ON
PSA SWON ¼ FS  VIN  IOUT  ð1:32Þ
2  ISA SWON

When power switch SA is turned off, internal parasitic capacitor CGS must dis-
charge until the voltage of capacitor CGS is equal to Miller plateau voltage VPLAT.
At this moment, Miller parasitic capacitor CGD is also discharged to equal Miller
plateau voltage VPLAT so that the voltage of capacitor CGS can be maintained for a
while. At the same time, voltage VLX is reduced to zero, and voltage VDS is
increased to input voltage VIN within T3 time (Fig. 1.27). When the voltage of
capacitor CGS is smaller than Miller plateau voltage VPLAT, current IDS is reduced to
zero until the voltage of capacitor CGS is equal to threshold voltage VTH within T4
time (Fig. 1.27). In Fig. 1.27, the turn off switching loss PSA_SWOFF of power
switch SA is equal to the power dissipation in the area during T1 and T2 time, as
shown in Eqs. (1.33)–(1.36).

T3 þ T4 TSA SWOFF
PSA SWOFF ¼ FS  VDS  IDS  ¼ FS  VIN  IOUT 
2 2
ð1:33Þ

QSW OFF
TSA SWOFF ¼ ð1:34Þ
ISA SWOFF

Fig. 1.27 Turn off switching A


loss PSA_SWOFF of the power
switch SA in the area during IDS IOUT
T3 time and T4 time

t
T3 T4
V

VDS VIN

0
t
VGS
VIN
VPLAT
VTH
t
QSW_OFF QTH
QGS QGD
28 1 Introduction

VPLAT  0
ISA SWOFF ¼ ð1:35Þ
RSOFF ON þ RGD þ RG

QSW OFF
PSA SWOFF ¼ FS  VIN  IOUT  ð1:36Þ
2  ISA SWOFF

Finally, the switching loss SA_SW of power switch SA can be computed as the
difference between turn on switching loss PSA_SWON and turn off switching loss
PSA_SWOFF by using Eq. (1.37).

PSA SW ¼ PSA SWON þ PSA SWOFF ð1:37Þ

Figure 1.28 shows the level shift driver circuit that drives power switch SB. This
level shift driver circuit is applied to change the power domain from IC internal
power to external input voltage VIN. Power switch SB has internal parasitic
capacitors, such as CGD, CGS, and CDS. Power switch SB is turned on by the voltage
of capacitor CGS, and voltage VDS is equal to the voltage drop of the body diode,
which is generally smaller than 0.7 V. For this reason, the turn on switching loss
PSB_SWON of power switch SB can be negligible. Meanwhile, power switch SB is
turned off by the voltage of capacitor CGS, and the output load current IOUT con-
tinues running in the same direction through the body diode. Thus, voltage VDS is
also small. For the same reason, the turn on switching loss PSB_SWON of power
switch SB can be negligible. Accordingly, the switching loss SB_SW of power switch
SB is minimal.
The switching loss of MOSFETs is equal to the sum of power dissipation
PSA_SW and power dissipation PSB_SW. Thus, the switching loss of MOSFETs
approximates power dissipation PSA_SW. The switching loss of MOSFETs gener-
ally occurs at a short time during the period when the power switches are turned on

VLX

Drain
VIN SB
CGD

SON
Turn-ON
RGD RG
Gate CDS
Duty Cycle SOFF Turn-OFF

CGS

Source

Fig. 1.28 The level shift driver circuit that drives power switch SB
1.4 Efficiency 29

and turned off. Hence, it exerts no effect when the output load current is heavy.
Moreover, when the output load current is light and the switching loss of MOSFETs
is continuous, load efficiency is affected. For this reason, the output load current is
light, and the switching loss of MOSFETs should be maintained. Therefore, the
switching loss of MOSFETs should be reduced to increase efficiency at light loads.
The PFM control mode is widely implemented to reduce the switching loss of
MOSFETs because the PFM control mode does not generate the same number of
control pulses like the PWM control mode does. At a light load, the PWM control
mode should force the duty cycle generation at the same switching frequency, so
the PWM control mode may result in a larger switching loss than the PFM control
mode. Figure 1.29 shows the control signals of the PWM control mode for buck
converter. For a brief analysis, Fig. 1.29 ignores the effects of output voltage ripple,
dead-time consideration, voltage drop of power switch SA, and voltage drop of
power switch SB. The minimum inductor current IL_MIN is a negative current, so the
PWM control mode allows a negative inductor current to waste energy. However,
the zero-current detection (ZCD) function [47–51] is widely used to prevent the
inductor current IL from becoming a negative current to achieve improved

VUG VIN

0
t

VLG VIN

0
t
VLX
VIN

0
t
A

IL IL_MAX
ΔIL
0
IL_MIN
t

Fig. 1.29 Control signals of the PWM control mode for buck converter
30 1 Introduction

efficiency at a light load. The ZCD function is usually implemented with the PWM
control mode or PFM control mode to improve efficiency at a light load.
The ZCD function can detect inductor current IL or voltage VLX to control power
switch SB. If the ZCD function detects inductor current IL, the ZCD function will
change the voltage signal VZCD from high to low level to turn off power switch SB
when inductor current IL is less than or equal to zero. The ZCD function can also
detect voltage VLX. When voltage VLX is greater than or equal to zero, the ZCD
function changes the voltage signal VZCD from high to low level to turn off power
switch SB. At this moment, inductor current IL goes through the body diode of
power switch SB, and inductor current IL equals zero quickly. Thus, voltage VLX is
equal to output voltage VOUT and not equal to zero voltage, such as the PWM
control mode without the ZCD function.
Figure 1.30 shows the control signals of the PWM control mode with the ZCD
function for buck converter. For a brief analysis, Fig. 1.30 ignores the effects of

VUG VIN

0
t

VLG VIN

0
t

VZCD High Level

Low Level
t
VLX
VIN

VOUT
0
t
A

IL
IL_MAX
ΔIL
IL_MIN (0)
t

Fig. 1.30 Control signals of the PWM control mode with the ZCD function for buck converter
1.4 Efficiency 31

output voltage ripple, dead-time consideration, voltage drop of power switch SA,
and voltage drop of power switch SB. The minimum inductor current IL_MIN is equal
to zero, so the PWM control mode with the ZCD function has no negative inductor
current to waste energy. The width of voltage VUG for the PWM control mode with
the ZCD function is according to the output load current, not to the ratio of the
output voltage to the input voltage. Thus, the width of voltage VUG for the PWM
control mode with the ZCD function is shorter than the duty cycle of the PWM
control mode. Once the voltage signal VZCD is changed from high to low level to
turn off power switch SB, the width of voltage VLG for the PWM control mode with
the ZCD function is according voltage signal VZCD. The inductor current ripple DIL
of the PWM control mode is larger than that of the PWM control mode with the
ZCD function because the inductor current IL of the PWM control mode with the
ZCD function has no negative current. Hence, the maximum inductor current
IL_MAX should be less than the PWM control mode without the ZCD function.
In general, the PFM control mode with the ZCD function not only prevents the
negative inductor current from wasting energy for power ICs but can also signifi-
cantly reduce the number of control pulses to increase light load efficiency.
Figure 1.31 shows the control signals of the PFM control mode with the ZCD
function for buck converter. For a brief analysis, Fig. 1.31 ignores the effects of
output voltage ripple, dead-time consideration, voltage drop of power switch SA,
and voltage drop of power switch SB. The minimum inductor current IL_MIN is equal
to zero, similar to the PWM control mode with the ZCD function. The width of
voltage VUG for the PWM control mode with the ZCD function is according to the
ratio of output voltage to input voltage, so the width of voltage VUG for the PFM
control mode with the ZCD function is equal to the duty cycle of the PWM control
mode. For this reason, the inductor current ripple DIL of the PFM control mode with
the ZCD function is larger than that of the PWM control mode. The inductor current
ripple DIL of the PFM control mode with the ZCD function is equal to that of the
PWM control mode because the width of voltage VUG for the PFM control mode
with the ZCD function is similar to that of the PWM control mode. Therefore, the
inductor current ripple DIL of the PFM control mode with the ZCD function is
similar to that of the PWM control mode. However, the minimum inductor current
IL_MIN for the PWM control mode is negative, so the PWM control mode cannot
save extra energy like the PFM control mode with the ZCD function does.
Once voltage signal VZCD is changed from high to low level to turn off power
switch SB, the width of voltage VLG for the PWM control mode with the ZCD
function is according to voltage signal VZCD. Unlike the PWM control mode, the
PFM control mode does not generate the same number with a fixed switching
frequency control, so the PFM control mode with the ZCD function can be useful in
reducing switching and conduction losses at light loads.
To further understand the advantages and superiority of the PFM control mode
with the ZCD function for buck converter to increase light load efficiency.
32 1 Introduction

VUG VIN

0
t

VLG VIN

0
t
VZCD
High Level

Low Level
t
VLX
VIN

VOUT
0
t
A

IL IL_MAX
ΔIL
IL_MIN (0)
t

Fig. 1.31 Control signals of the PFM control mode with the ZCD function for buck converter

Experimental verifications were conducted to compare the feasibility and perfor-


mance between the PFM control mode with the ZCD function and the PWM control
mode for buck converter. To achieve a fair and correct comparison between the PFM
control mode with the ZCD function and the PWM control mode for buck converter,
so the experimental results are measured by the same operation conditions.
The specifications are as follows:
(1) Input voltage (VIN): 12 V
(2) Output voltage (VOUT): 1.05 V
(3) Output load current (IOUT): 10 mA
(4) Switching frequency (FS): 630 kHz
1.4 Efficiency 33

Fig. 1.32 Comparison of the VLX (10V/4μs)


experimental results between
the PFM control mode with (1)
the ZCD function and the IL (500mA/4μs)
PWM control mode at the
light load for buck converter (2)

VOUT
VLX (10V/4μs)
(3)
0V

IL (500mA/4μs)
(4)

(1), (2) PWM Mode Control without ZCD Function


(3), (4) PFM Mode Control with ZCD Function

(5) MOSFET (SA, SB): BSC0909NS * 3


(6) Feedback resistors (RD1, RD2): 13.2 kX and 12 kX
(7) Main inductor (L): IHLP4040DZER1R0MA1 (1 lH)
(8) Output capacitors (CO): 22 lF/6.3 V (RCO: 3 mX)*2
(9) Reference voltage (VREF): 0.5 V
Figure 1.32 compares the experimental results between the PFM control mode
with the ZCD function and the PWM control mode at the light load for buck
converter. Figure 1.32 represents the VLX voltage and the IL current the PFM
control mode with the ZCD function and the PWM control mode at the light load.
In Fig. 1.32, the red-colored waveforms represent the VLX voltage and the IL cur-
rent for the PWM control mode. The blue-colored waveforms represent the VLX
voltage and the IL current for the PFM control mode with the ZCD function. The
signals are measured based on the same output load current. The switching fre-
quency of the PWM control mode maintains 630 kHz at output load current
10 mA. The switching frequency of the PFM control mode with the ZCD function
is based on the output load current condition and the switching frequency is
measured 33.3 kHz at output load current 10 mA. The number of control pulses for
the PWM control mode is larger than the PFM control mode with the ZCD func-
tion, so the PWM control mode has a large switching loss. In addition, the IL
current of the PFM control mode with the ZCD function is maintained zero current
during 28 ls, so it is useful to save conduction loss on the SB switch, because the
inductor current is through the body diode of the SB switch. Moreover, the DIL
ripple current is 1 A of the PFM control mode with the ZCD function and it is
similar to the DIL ripple current of the PWM control mode with the same width of
the voltage signal VUG. Based on these reasons, the PFM control mode with the
ZCD function can significantly achieve a better efficiency at the light load.
34 1 Introduction

References

1. P. Burger, Analysis of a class of pulse modulated dc-to-dc power converters, IEEE Trans. Ind.
Elect. Contr. Instrum. IECI-22, 104 (1975)
2. L.E. Gallaher, Current regulator with AC and DC feedback, U.S. Patent 3 350 628, 31 Oct
1967
3. A.D. Schoenfeld, Y. Yu, ASDTIC control and standardized interface circuits applied to buck,
parallel and buck-boost DC-to-DC power converters (NASA, Washington, DC, NASA
Rep. NASA CR-121106, Feb. 1973)
4. C.W. Deisch, Switching control method changes power converter into a current source, in
Proceedings IEEE Power Electronics Specialists Conference, (1978), pp. 300–306
5. P.L. Hunter, Converter circuit and method having fast responding current balance and
limiting, U.S. Patent 4 002 963, 1 Nov 1977
6. L.H. Dixon, Average current-mode control of switching power supplies, in Proceedings
Unitrode Power Supply Design Seminars Handbook, (1990), pp. 5.1–5.14
7. N. Mohan, Power electronics circuits: An overview,” in Proceedings IEEE Industrial
Electronics Society Conference, (1988), pp. 522–527
8. N. Mohan, W.P. Robbins, P. Imbertson, T.M. Undeland, R.C. Panaitescu, A.K. Jain, P. Jose,
T. Begalke, Restructuring of first courses in power electronics and electric drives that
integrates digital control. IEEE Trans. Power Electron. 18, 429–437 (2003)
9. R.D. Middlebrook, S. Cuk, A general unified approach to modeling switching-converter
power states, in Proceedings of IEEE Power Electronics Specialists Conference, (1976),
pp. 18
10. D.Y. Chen, H.A. Owen, T.G. Wilson, Computer aided design and graphics applied to the
study of inductor-energy-storage dc-to-dc electronic power converters. IEEE Trans.
Aerosp. Electron. Syst. AES-9, 585 (1973)
11. R.W. Erickson, D. Maksimovic, Fundamentals of power electronics (Norwell, MA, Kluwer,
2001)
12. K.H. Chen, H.W. Huang, S.Y. Kuo, Fast-transient DC–DC converter with on-chip
compensated error amplifier. IEEE Trans. Circuits Syst. II Express Briefs 54(12), 1150–
1154 (2007)
13. C.L. Chiu, K.H. Chen, A high accuracy current-balanced control technique for LED
backlight, in Proceendings of IEEE Power Electronics Specialists Conference, (2008),
pp. 4202–4206
14. D. Rand, B. Lehman, A. Shteynberg, Issues, models and solutions for triac modulated phase
dimming of led lamps, in 2007 IEEE Power Electronics Specialists Conference, (June 2007),
pp. 1398–1404
15. A. Morrison, J.W. Zapata, S. Kouro, M.A. Perez, T.A. Meynard, H. Renaudineau, Partiat
power dc-dc converter for photovoltaic two-stage string inverters, in 2016 IEEE Energy
Conversion Congress and Exposition (ECCE), (September 2016)
16. J.W. Zapata, T.A. Meynard, S. Kouro, Partial power dc-dc converter for large-scale
photovoltaic systems, in 2016 IEEE 2nd Annual Southern Power Electronics Conference
(SPEC), (December 2016), pp. 1–6
17. S. Saponara, G. Pasetti, N. Costantino, F. Tinfena, P.D’Abramo, L. Fanucci, A flexible LED
driver for automotive lighting applications: IC design and experimental characterization. IEEE
Trans. Power Electron. 27(3), 1071–1075 (2012)
18. C.-S. Moo, Y.-J. Chen, W.-C. Yang, An efficient driver for dimmable LED lighting. IEEE
Trans. Power Electron. 27(11), 4613–4618 (2012)
19. G. Sauerlander, D. Hente, H. Radermacher, E. Waffenschmidt, J. Jacobs, Driver electronics
for LEDs, Proc. Conf. Rec. IEEE Ind. Appl. Conf. 2621–2626 (2006)
20. G. Rincon-Mora, P. Allen, A low-voltage low quiescent current low drop-out regulator.
IEEE J. Solid-State Circuits. 33(1), 36–44 (1998)
References 35

21. R.J. Milliken, J. Silva-Martínez, E. Sánchez-Sinencio, Full On-Chip CMOS Low-Dropout


Voltage Regulator. IEEE Trans. Circuits Syst. I. Reg. Pap. 54(9), 1879 (2007)
22. C. Chen, J. Wu, Z. Wang, 150 mA LDO with self-adjusting frequency compensation scheme,
Electron. Lett. 47(13), 767–768 (2011)
23. A. Patel, G. Rincon-Mora, High Power-Supply-Rejection (PSR) Current-Mode Low-Dropout
(LDO) Regulator. IEEE Trans. Circuits Syst. II Exp. Briefs. 57(11), 868–873 (2010)
24. E.N.Y. Ho, P.K.T. Mok, A capacitor-less CMOS active feedback low-dropout regulator with
slew-rate enhancement for portable on-chip application. IEEE Trans. Circuits Syst. II
Exp. Briefs. 57(2), 80-84 (2010)
25. K.N. Leung, P. Mok, A capacitor-free CMOS low-dropout regulator with damping-
factor-control frequency compensation. IEEE J. Solid-State Circuits. 38(10), 1691–1702
(2003)
26. H. Lee, P. Mok, K.N. Leung, Design of low-power analog drivers based on slew-rate
enhancement circuits for cmos low-dropout regulators. IEEE Trans. Circuits Syst. II
Exp. Briefs. 52(9), 563–567 (2005)
27. J.R. Huang, C.H. Wang, C.J. Lee, K.L. Tseng, D. Chen, Native AVP control method for
constant output impedance of DC power converters, in Proceedings of IEEE Power
Electronics Specialists Conference, (2007), pp. 2023–2028
28. A. Waizman, C.Y. Chung, Resonant free power network design using extended adaptive
voltage positioning (EAVP) methodology. IEEE Trans. Adv. Packag. 24, 236–244 (2001)
29. M. Lee, D. Chen, K. Huang, E. Tseng, B. Tai, Compensator design for adaptive voltage
position (AVP) for multiphase VRMs, in Proceedings of IEEE Power Electronics Specialists
Conference, (2006)
30. K. Yao, Y. Meng, P. Xu, F.C. Lee, Design considerations for VRM transient response based
on the output impedance, in Proceedings of IEEE Applied Power Electronics Conference and
Exposition conference, (2002), pp. 14–20
31. P.L. Wong, Performance improvements of multi-channel interleaving voltage regulator
modules with integrated coupling inductors, Dissertation, Virginia Polytechnic Institute and
State University, 2001
32. S.K. Mishra, Design-oriented analysis of modern active droop controlled power supplies.
IEEE Trans. Ind. Electron. 56(9), 3704–3708 (2009)
33. J.A.A. Qahouq, V. Arikatla, Power converter with digital sensorless adaptive voltage
positioning control scheme. IEEE Trans. Ind. Electron. 58(9), 4105–4116 (2010)
34. W.W. Chen, J.F. Chen, T.J. Liang, Dynamic Ramp control in current-mode adaptive on-time
control for Buck converter on chip, in Proceedings of IEEE Future Energy Electronics
Conference and ECCE Asia (IFEEC 2017 - ECCE Asia), (2017), pp. 280–285
35. W.W. Chen, J.F. Chen, T.J. Liang, J.R. Huang, W.Y. Ting, Improved transient response using
HFFC in current-mode CFCOT control for buck converter, in Proceedings IEEE
International Conference on Power Electronics and Drive Systems (PEDS), (2013),
pp. 546–549
36. W.W. Chen, J.F. Chen, T.J. Liang, J.R. Huang, L.C. Wei, W.Y. Ting, Implementing dynamic
quick response with high-frequency feedback control of the deformable constant on-time
control for Buck converter on-chip. IET Power Electron. 6(4), 383–391 (2013)
37. Richtek Tech. Corp., Analysis of Buck Converter Efficiency, Application Note, (2014)
38. J. Gallaghe, Coupled inductors improve multiphase buck efficiency. Power Electron. Technol.
(2006)
39. Fairchild Semiconductor Inc., AN-6005 Synchronous buck MOSFET loss calculations with
Excel model. Application Note, 2014
40. K. Yao, Y. Meng, F.C. Lee, A novel winding coupled-buck converter for high-frequency high
step-down DC/DC conversion, in Proceedings IEEE Power Electronics Specialists
Conference, (2002)
36 1 Introduction

41. H. Krishnamurthy, V. Vaidya, P. Kumar, G. Matthew, S. Weng, B. Thiruvengadam, W.


Proefrock, K. Ravichandran, V. De, A 500 MHz 68% efficient fully on-die digitally
controlled buck voltage regulator on 22 nm tri-gate CMOS, in Symposium VLSI Circuits
Digest Technical Papers, (2014), pp. 210–211
42. S.J. Kim, Q. Khan, M. Talegaonkar, A. Elshazly, A. Rao, N. Griesert, G. Winter, W.
McIntyre, P. Hanumolu, High frequency buck converter design using time-based control
techniques. IEEE J. Solid-State Circuits. 50(4), 990–1001 (2015)
43. W.R. Liou, M.L. Yeh, Y.L. Kuo, A high efficiency dual-mode buck converter IC for portable
applications, IEEE Trans. Power Electron. 23(2), 667–677 (2008)
44. Texas Instruments Inc., MOSFET power losses and how they affect power-supply efficiency,
Application Report, 2016
45. Texas Instruments Inc., Calculating Efficiency, Application Report, 2010
46. Texas Instruments Inc., Optimizing MOSFET characteristics by adjusting Gate Drive
Amplitude, Application Report, 2005
47. A.V. Petershevs, S.R. Sanders, Digital multimode buck converter control with
loss-minimizing synchronous rectifier adaptation. IEEE Trans. Power Electron. 21(6),
1588–1599 (2006)
48. S. Angkititrakul, H. Hu, Design and analysis of buck converter with pulse-skipping
modulation, in Proceedings of IEEE Power Electronics Specialists Conference, (2008),
1151–1156
49. X. Zhou, M. Donati, L. Amoroso, F.C. Lee, Improved light-load efficiency for synchronous
rectifier voltage regulator module. IEEE Trans. Power Electron. 15(5), pp. 826–834 (2000)
50. C.L. Chen, W.L. Hsieh, W.J. Lai, K.H. Chen, C.S. Wang, A new PWM/PFM control
technique for improving efficiency over wide load range, in Proceedings of IEEE
International Conference on Electronics, Circuits and Systems, (2008), pp. 962–965
51. S. Kapat, S. Banerjee, A. Patra, Discontinuous map analysis of a DC-DC converter governed
by pulse skipping modulation. IEEE Trans. Circuits System. Part I. 57(7), 1793–1801 (2010)
Chapter 2
Review of the PWM Control Circuits
for Power Converters

2.1 Voltage-Mode Control Circuit for Power Converters

Power converters are electrical control circuits that transfer energy from a DC
voltage source to the output loading and regulate the output voltage to meet the user
design. Energy is transferred via electronic switches made with transistors and
diodes to an output filter and then transferred to the output loading.
These converters employ square-wave PWM to achieve voltage regulation. The
output voltage is regulated by varying the duty cycle of the power semiconductor
switch driving signal. The voltage waveform across the switch and at the input of
the filter is a square wave in nature and generally results in high switching losses
when the switching frequency is increased. However, these converters are easy to
control, are well understood, and have a wide load control range. These converters
also operate at a fixed-frequency, variable duty cycle. This type of control signal is
called PWM control signal. Depending on the duty cycle, these converters can
operate in either CCM or discontinuous conduction mode (DCM). If the current
through the output inductor never reaches zero, then the converter operates in CCM.
If the current through the output inductor reaches zero, then the converter operates
in DCM.
The output voltage is equal to the average value in the switching cycle of the
voltage applied at the output filter. For real switches with parasitic elements, effi-
ciency depends on conduction and switching losses, but the efficiency of power
converters remains higher than that of linear regulators such as LDO.
Power converters are widely applied in portable electronic equipment and
products, especially those designed to reduce standby power loss. They demonstrate
high efficiency and present a fast transient response due to system design. The
fixed-frequency PWM control scheme for power converters [1–24] is commonly
used with current-mode and voltage-mode control circuits.
Voltage-mode control circuit is the simplest circuit structure for PWM control
scheme for power converters. The major characteristic of this design is the presence

© Springer Nature Singapore Pte Ltd. 2018 37


W.-W. Chen and J.-F. Chen, Control Techniques for Power
Converters with Integrated Circuit, Power Systems,
https://doi.org/10.1007/978-981-10-7004-4_2
38 2 Review of the PWM Control Circuits for Power Converters

Fig. 2.1 Circuit diagram of S1 VLX L


the voltage-mode control VIN VOUT
circuit for buck converter VUG IL
VLG S2 RCO
IOUT
CO
RD1
Driver Ramp
Generator

VRAMP
VFB
VCOMP
GM

RC RD2
VREF

CC

of a single voltage feedback path, with PWM performed by comparing the voltage
error signal with a constant ramp waveform. Current limiting must be conducted
separately. The advantages of voltage-mode control are as follows: the single
feedback loop is easy to design and analyze and a large-amplitude ramp waveform
provides good noise immunity for a stable modulation process. Figure 2.1 shows
the circuit diagram of the voltage-mode control circuit for buck converter. S1 and S2
are the power switches integrated on-chip, L is the output inductor. RCO is the ESR
of output capacitor CO. Current source IOUT is the output load current. The driver
circuit uses the input signal on-time width to generate two control signals VUG and
VLG, these two signals should be avoid to turn on at the same time, because this
operation make this system to have a shoot through problem. The compensator of
RC and CC should be designed an optimization to increase the transient response.
Only the feedback signal VFB and reference voltage VREF are built inside the IC.
The output signal of comparator depends on the input signals VCOMP and VRAMP
results.
The PWM three-terminal model [23, 24] is a good tool to analyze loop stability
because power switches S1 and S2 can be modeled as equivalent circuits similar to a
dependent voltage source, a dependent current source, and an ideal transformer.
The PWM three-terminal model with voltage-mode control circuit for the buck
converter is shown in Fig. 2.2. In Fig. 2.2, a transfer function for control-to-output
GCTO(s) is obtained with Eqs. (2.1)–(2.7). Based on the GCTO(s) transfer function,
the voltage-mode control circuit for the buck converter has one zero and two poles
system, as shown in Fig. 2.3. The DC gain of control-to-output depends on the
input voltage by Eq. (2.2), and a large input voltage has a large DC gain. The
voltage-mode control circuit for the buck converter is unsuitable for a wide input
voltage range. The zero SZ1 depends on output capacitor CO and ESR RCO of the
output capacitor. With regard to output capacitor selection, tantalum capacitors are
2.1 Voltage-Mode Control Circuit for Power Converters 39

Fig. 2.2 PWM VIN


three-terminal model with d
voltage-mode control circuit D L
a c
for buck converter
iL i OUT
I Ld 1 D RCO
ROUT VOUT
CO
p

d
Fm

FC VCOMP

Fig. 2.3 Control-to-output


with one zero and two poles dB Fm·VIN
system of voltage-mode Q
control circuit for buck
converter
O

-40dB/dec

-20dB/dec
Sz1
rad/s

a poor choice for output capacitors because tantalum capacitors usually fail to create
a short circuit across their terminals, thereby raising the possibility of a fire hazard.
Ceramic or aluminum electrolytic capacitors are preferred because they do not have
this failure mode. Ceramic capacitors possessing small footprint, low profile, low
ESR, low cost, and high reliability are widely used in microprocessor decoupling
and power converter filtering applications. Thus, ceramic capacitors are a good
choice when the evaluation board area, cost reduction, or component height is
considered. The low ESR of ceramic capacitors results in high performance and
efficiency but may cause the system loop to become unstable. Generally, the ESR of
a ceramic output is less than 10 mX, and the SZ1 zero location is set at a high
frequency of the voltage-mode control circuit for the buck converter. Thus, the
control-to-output GCTO(s) is changed to a two poles system.
40 2 Review of the PWM Control Circuits for Power Converters

V^OUT 1 þ s=SZ1
GCTO ðsÞ ¼ ¼ Fm  VIN ð2:1Þ
^COMP
V 1 þ s=ðxO  QÞ þ s2 =x2O

GCTO ð0Þ ¼ Fm  VIN ð2:2Þ

1
xO ¼ pffiffiffiffiffiffiffiffiffiffiffiffi ð2:3Þ
L  CO
rffiffiffiffiffiffi
CO
Q ¼ ROUT  ð2:4Þ
L
1
Fm ¼ ð2:5Þ
VRAMP

FC ¼ 1 ð2:6Þ

1
SZ1 ¼ ð2:7Þ
CO  RCO

Table 2.1 lists the operation conditions of the voltage-mode control circuit for
the buck converter. According to Table 2.1, the comparison of MathCAD predic-
tions and SIMPLIS simulation results of the open-loop control-to-output bode plot
for the buck converter is plotted in Fig. 2.4. In Fig. 2.4, the MathCAD predictions
verify the SIMPLIS simulation results. The red-colored line represents the
MathCAD predictions, and the blue-colored dot represents the SIMPLIS simulation
results. The DC gain curve of the open-loop control-to-output bode plot is equal to
6.02 dB as obtained by Eq. (2.2). The two poles are located at 2.475 kHz by
Eq. (2.3). The two poles cause a sharp phase drop of 180°, so the voltage-mode
control circuit requires the addition of one zero to cancel the effect of the two poles.
Based on these operation conditions using a large output capacitor, the zero is
located at 18.09 kHz, so it can help increase the phase degree (30°). In general, the
output capacitor uses a ceramic capacitor with high switching frequency, and the
capacitance of a single capacitor should be less than 50 lF. Thus, the zero is mainly
located at more than 300 kHz with 10 mX of ESR. The zero at high frequency
cannot help increase the phase degree.
A buck converter can achieve step-down voltage from its input power supply to
its output terminal, so it is also widely used to convert a computer’s main supply

Table 2.1 Operation conditions in voltage-mode control circuit for buck converter
VIN (V) 4.8 V FS (kHz) 500 kHz L (lH) 4.7 lH
VOUT (V) 1.2 V RD1 (kX) 100 kX CO (lF) 880 lF
IOUT (A) 5A RD2 (kX) 40 kX RCO (mX) 10 mX
VREF (V) 0.8 V VRAMP (V) 2.4 V
2.1 Voltage-Mode Control Circuit for Power Converters 41

(a) Gain

(b) Phase

Fig. 2.4 MathCAD predictions of open-loop control-to-output bode plot for buck converter

voltage (often 12 V) down to lower voltages needed by USB, DRAM, and the CPU
(1.8 V or less), or the output voltage is designed to be lower than the input power
supply. A boost converter is different from a buck converter because a boost
converter achieves step-up voltage from its input power supply to its output ter-
minal; hence, it is widely used to convert a power supply voltage up to larger
voltages needed by the display power driver and LED driver, or the output voltage
is designed to be larger than input power supply.
Figure 2.5 shows a circuit diagram of the voltage-mode control circuit for boost
converter. S1 and S2 are the power switches integrated on-chip, L is the output
inductor. RCO is the ESR of output capacitor CO. Current source IOUT is the output
load current. The driver circuit uses the input signal on-time width to generate two
control signals VUG and VLG, these two signals should be avoid to turn on at the
same time, because this operation make this system to have a shoot through
problem. The compensator of RC and CC should be designed an optimization to
42 2 Review of the PWM Control Circuits for Power Converters

Fig. 2.5 Circuit diagram of L VLX S1


the voltage-mode control VIN VOUT
circuit for boost converter IL VUG
VLG S2 RCO
VUG IOUT
CO
RD1
Driver Ramp
Generator

VRAMP
VFB
VCOMP
GM

RC RD2
VREF

CC

increase the transient response. Only the feedback signal VFB and reference voltage
VREF are built inside the IC. The output signal of comparator depends on the input
signals VCOMP and VRAMP results.
The PWM three-terminal model [23, 24] with the voltage-mode control circuit
for the boost converter is shown in Fig. 2.6. In Fig. 2.6, a transfer function for

Fig. 2.6 PWM L c


three-terminal model with
voltage-mode control circuit D
for boost converter iL i OUT
p R
CO
1
-VOUT ROUT V
OUT
d
D I Ld CO

d
Fm

FC VCOMP
2.1 Voltage-Mode Control Circuit for Power Converters 43

Fig. 2.7 Control-to-output 2


with one zero, one RHP zero, dB Fm·(VIN/(1- D) )
and two poles system of Q
voltage-mode control circuit
for boost converter
O -40dB/dec

-20dB/dec
Sz2 0dB/dec

Sz2

rad/s

control-to-output GCTO(s) can be obtained by Eqs. (2.8)–(2.16). Based on the


GCTO(s) transfer function, the voltage-mode control circuit for the boost converter
has one zero, one right-half-plane (RHP) zero, and two poles system as shown in
Fig. 2.7. The DC gain of control-to-output depends not only on the input voltage
but also on the duty cycle according to Eq. (2.9). A large duty cycle has a large DC
gain. Given this effect, the voltage-mode control circuit for the boost converter is
unsuitable for generating very large output voltages. For example, if the duty cycle
is 0.9, the DC gain of control-to-output should be increased to 40 dB with the same
input voltage. A large DC gain of control-to-output makes it difficult to design an
optimal compensator to ensure loop system stability. The first zero SZ1 depends on
output capacitor CO and ESR RCO of the output capacitor. Ceramic capacitors
possessing a small footprint, low profile, no failure mode, low ESR, low cost, and
high reliability are widely used in microprocessor decoupling and power converter
filtering applications. The low ESR of ceramic capacitors for output capacitors
results in high performance and efficiency but may cause the system loop to become
unstable. In general, the ESR of the ceramic output is less than 10 mX, and the first
zero SZ1 location is set at a high frequency of the voltage-mode control circuit for
the boost converter. Thus, the control-to-output GCTO(s) is changed to two poles
and one RHP zero system. The second zero SZ2 is an RHP zero. RHP zero has the
same 20 dB/decade rising gain magnitude as a conventional zero, but with a 90°
phase lag instead of lead. This characteristic is difficult, if not impossible, to
compensate. The designer is usually forced to roll off the loop gain at a relatively
low frequency. The crossover frequency may be a decade or more below what it
otherwise could be, resulting in severe impairment of the dynamic response. In
general, RHP zero is set at a frequency higher than two poles for the boost con-
verter. The RHP zero phase drop starts a decade earlier and negatively affects the
potential phase margin of the converter’s control loop. This is the nature of
instability of a voltage-mode controlled boost converter running in CCM.
44 2 Review of the PWM Control Circuits for Power Converters

V^OUT VIN ð1 þ s=SZ1 Þ  ð1  s=SZ2 Þ


GCTO ðsÞ ¼ ¼  Fm  ð2:8Þ
^COMP ð1  DÞ
V 2 1 þ s=ðxO  QÞ þ s2 =x2O

VIN
GCTO ð0Þ ¼  Fm ð2:9Þ
ð1  DÞ2

1
xO ¼ pffiffiffiffiffiffiffiffiffiffiffiffiffiffiffi ð2:10Þ
LD  CO
rffiffiffiffiffiffi
CO
Q ¼ ROUT  ð2:11Þ
LD

1
Fm ¼ ð2:12Þ
VRAMP

FC ¼ 1 ð2:13Þ

1
SZ1 ¼ ð2:14Þ
CO  RCO

ROUT
SZ2 ¼ ð2:15Þ
LD
L
LD ¼ ð2:16Þ
ð1  D Þ2

Table 2.2 lists the operation conditions of the voltage-mode control circuit for
the boost converter. According to Table 2.2, the comparison of MathCAD pre-
dictions and SIMPLIS simulation results of the open-loop control-to-output bode
plot for the boost converter is plotted in Fig. 2.8. In Fig. 2.8, the MathCAD pre-
dictions verify the SIMPLIS simulation results. The red-colored line represents the
MathCAD predictions, and the blue-colored dot represents the SIMPLIS simulation
results. The DC gain curve of the open-loop control-to-output bode plot is equal to
25.666 dB according to Eq. (2.9). The two poles are located at 6.741 kHz by
Eq. (2.10). The two poles cause a sharp phase drop of 180°, so the voltage-mode
control circuit requires the addition of one zero to cancel the effect of the two poles.
Based on this operation condition using the 44 lF output capacitor, the first zero
SZ1 is located at 361.7 kHz by Eq. (2.14), and the second zero SZ2 is located at

Table 2.2 Operation conditions in voltage-mode control circuit for boost converter
VIN (V) 5V FS (kHz) 1000 kHz L (lH) 2.2 lH
VOUT (V) 12 V RD1 (kX) 120 kX CO (lF) 44 lF
IOUT (A) 0.3 A RD2 (kX) 8.57 kX RCO (mX) 10 mX
VREF (V) 0.8 V VRAMP (V) 1.5 V
2.1 Voltage-Mode Control Circuit for Power Converters 45

(a) Gain

(b) Phase

Fig. 2.8 Comparison of MathCAD predictions and SIMPLIS simulation results open-loop
control-to-output bode plot for boost converter

502 kHz by Eq. (2.15). Thus, the first zero SZ1 can cancel the effect of the second
zero SZ2. The phase of open-loop control-to-output maintains a two-pole behavior
similar to a sharp phase drop of 180°. Meanwhile, the RHP zero SZ2 depends on the
output loading and duty cycle, so the user needs to consider an optimal compen-
sator at the worst conditions to ensure that the RHP zero SZ2 does not affect the
system loop stability.

2.2 Current-Mode Control Circuit for Power Converters

The current-mode control circuit contains two feedback control signals and differs
from the voltage-mode control circuit. The output voltage is fed to an error amplifier
to generate control signal VCOMP. Inductor current IL is sampled into voltage signal
46 2 Review of the PWM Control Circuits for Power Converters

Fig. 2.9 Circuit diagram of S1 VLX L VOUT


the current-mode control
circuit for buck converter VIN VUG IL
VLG RCO
S2 IOUT
Ri CO
RD1
VCS
Driver VRAMP Ramp
Generator
VSUM
VCOMP VFB
GM

RC VREF RD2

CC

VCS, and voltage signal VRAMP is added for comparison with control signal VCOMP to
control the peak inductor current and regulate output voltage VOUT to the specified
level. The inductor current IL feedback control signal causes the current-mode
control circuit to have a better transient response than the voltage-mode control
circuit. The current-mode control circuit shows additional advantages of improved
closed-loop stability and fast transient response [2–5, 8–22].
Figure 2.9 shows a circuit diagram of the current-mode control circuit for the
buck converter. Figure 2.9 shows the circuit diagram of the current-mode control
circuit for buck converter. S1 and S2 are the power switches integrated on-chip, L is
the output inductor. Ri is the current sense gain. RCO is the ESR of output capacitor
CO. Current source IOUT is the output load current. The driver circuit uses the input
signal on-time width to generate two control signals VUG and VLG, these two signals
should be avoid to turn on at the same time, because this operation make this
system to have a shoot through problem. The compensator of RC and CC should be
designed an optimization to increase the transient response. Only the feedback
signal VFB and reference voltage VREF are built inside the IC. The rising slope of
inductor current IL is sensed and required in the feedback loop as it contains supply
information for the current-mode control circuit. The ramp generator circuit that can
generate voltage signal VRAMP to be added to inductor current IL is sampled. The
sum voltage of voltage signal VRAMP and voltage signal VCS is used through a
resistor to obtain voltage signal VSUM. The output signal of the comparator depends
on the results of input signals VCOMP and VSUM.
However, the current-mode control circuit produces a subharmonic issue [2–5,
14–22, 25–27] when the duty cycle is more than 50%, as shown in Fig. 2.10.
Figure 2.10a shows that when the duty cycle is more than 50%, the duty cycle
cannot maintain the same width of control signal VUG, and a subharmonic issue
with control signal VUG arises. Figure 2.10b shows that when the duty cycle is less
than 50%, the duty cycle can maintain the same width of control signal VUG, so no
subharmonic issue arises with control signal VUG. The subharmonic issue can
2.2 Current-Mode Control Circuit for Power Converters 47

Fig. 2.10 Without the ramp V


compensation of Unstable : Duty Cycle > 50%
current-mode control circuit VCOMP
for buck converter
VCS

t
VUG

t
Ts 2Ts 3Ts 4Ts

(a) Duty cycle more than 50%


V
Stable : Duty Cycle < 50%
VCOMP

VCS

t
VUG

t
Ts 2Ts 3Ts 4Ts

(b) Duty cycle less than 50%

generate a different width of the duty cycle and may affect the output voltage ripple.
An output voltage ripple with a subharmonic issue is larger than without a sub-
harmonic issue.
To overcome the subharmonic issue, the ramp compensation signal is always
added to the control loop to prevent subharmonic, as shown in Fig. 2.11. Ramp
compensation is designed according to the inductor current ripple determined by the
input voltage, output voltage, and inductance. A general power control IC often
embeds a fixed ramp into the control loop. However, different operation conditions

Fig. 2.11 With the ramp V


compensation of
current-mode control circuit VCOMP Stable : Duty Cycle > 50%
VRAMP
for buck converter during
duty cycle more than 50%

VCS

t
VUG

t
Ts 2Ts 3Ts 4Ts
48 2 Review of the PWM Control Circuits for Power Converters

Fig. 2.12 PWM


three-terminal model with
current-mode control circuit
for buck converter

for input and output voltages can result in a varied inductor current ripple. Meeting
the fixed ramp compensation requires that the inductance be changed to maintain
the same current ripple. Thus, the compensation circuit should also be modified to
keep the same system loop stable.
The PWM three-terminal model [23, 24] with the current-mode control circuit
for the buck converter is shown in Fig. 2.12. In Fig. 2.12 a transfer function for
control-to-output GCTO(s) can be obtained by Eqs. (2.17)–(2.28). FP(s) provides
one pole and one zero [14–22]. The pole of FP(s) is located at the dominant
low-frequency characteristic of the system in Eqs. (2.19)–(2.20). FH(s) provides
two poles at half the switching frequency with good ramp compensation design in
Eqs. (2.22)–(2.24). The high-frequency effects are ignored. However, if the mC
value is small, it presents no ramp compensation or small ramp compensation.
A small mC value has two poles, which can cause a high Q at half of the switching
frequency. If the ramp compensation is large, it presents large ramp compensation.
At a large mC value, given that the ramp compensation is much larger than the
inductor current, the control loop is formed through a voltage-mode control
behavior. The current-mode control is more stable than the voltage-mode control
with only one dominant pole.

V^OUT ROUT 1
GCTO ðsÞ ¼ ¼  h  i  FpðsÞ  FhðsÞ
^COMP
V Ri 1 þ ROUT
1þ SE
 D0  0:5
LFS SN

ð2:17Þ
2.2 Current-Mode Control Circuit for Power Converters 49

ROUT 1
GCTO ð0Þ ¼  h  i ð2:18Þ
Ri 1 þ ROUT
1 þ SSNE  D0  0:5
LFS

1 þ s=SZ1
FpðsÞ ¼ ð2:19Þ
1 þ s=SP
  
1 1 SE
SP ¼ þ  1þ  D0  0:5 ð2:20Þ
CO  ROUT CO  L  FS SN

1
SZ1 ¼ ð2:21Þ
CO  RCO

1
FhðsÞ ¼ s þ s2 ð2:22Þ
1þ xn QP x2 n

1
QP ¼ h  i ð2:23Þ
p  1þ SE
SN  D0  0:5

xn ¼ p  FS ð2:24Þ

FC ¼ 1 ð2:25Þ

SE
mC ¼ 1 þ ð2:26Þ
SN

VIN  VOUT
SN ¼  Ri ð2:27Þ
L
SE ¼ VRAMP  FS ð2:28Þ

The DC gain of control-to-output directly depends on the output loading and


current sense gain in Eq. (2.18). The DC gain of control-to-output of the
current-mode control circuit is different from that of the voltage-mode control
circuit because the former is not directly dependent on the input voltage. Thus, the
current-mode control circuit is suitable for a wide range of input voltages.
Moreover, the current-mode control circuit has current feedback loop control, so the
DC gain is highly related to the current information on output loading and current
sense gain.
The zero SZ1 depends on output capacitor CO and the ESR RCO of the output
capacitor in Eq. (2.21). Low ESR of a ceramic capacitor for output capacitors is less
than 10 mX, and the SZ1 zero location is set at a high frequency of the current-mode
control circuit for the buck converter. The pole SP of control-to-output is the
dominant low frequency characteristic of the system in Eq. (2.20). It is different
from the voltage-mode control circuit because the inductor current IL feedback
50 2 Review of the PWM Control Circuits for Power Converters

Table 2.3 Operation conditions in current-mode control circuit for buck converter
VIN (V) 5V FS (kHz) 380 kHz L (lH) 15 lH
VOUT (V) 3.3 V RD1 (kX) 125 kX CO (lF) 22 lF
IOUT (A) 1A RD2 (kX) 40 kX RCO (mX) 10 mX
VREF (V) 0.8 V Ri (A/V) 0.4 A/V VRAMP (V) 0.358 mV

control signal can be similar to a zero behavior, so the current-mode control circuit
is approximately one pole less than the voltage-mode control circuit.
Table 2.3 lists the operation conditions of the current-mode control circuit for
the buck converter. Based on Table 2.3, a comparison of MathCAD predictions and
SIMPLIS simulation results of the open-loop control-to-output bode plot for the
buck converter is provided in Fig. 2.13. In Fig. 2.13, the MathCAD predictions
verify the SIMPLIS simulation results. The red-colored line represents the

(a) Gain

(b) Phase

Fig. 2.13 MathCAD predictions of open-loop control-to-output bode plot for buck converter
2.2 Current-Mode Control Circuit for Power Converters 51

Fig. 2.14 Circuit diagram of L VLX S1 VOUT


VIN
the current-mode control
circuit for boost converter IL VUG IL
RCO IOUT
VLG S2
VUG Ri CO
RD1
VCS
Driver VRAMP Ramp
Generator
VSUM
VCOMP VFB
GM

RC VREF RD2

CC

MathCAD predictions, and the blue-colored dot represents the SIMPLIS simulation
results. The DC gain curve of the open-loop control-to-output bode plot is equal to
14.819 dB by Eq. (2.18). The pole SP of control-to-output is located at 3.284 kHz
by Eq. (2.20). Based on this operation condition using the small output capacitor,
the zero is located at 723.4 kHz, so it cannot help increase the phase degree.
However, two poles of FH(s) exert an effect on the phase degree in these operation
conditions, and the phase degree is −140° at 100 kHz.
Figure 2.14 shows a circuit diagram of the current-mode control circuit for the
boost converter. S1 and S2 are the power switches integrated on-chip, L is the output
inductor. Ri is the current sense gain. RCO is the ESR of output capacitor CO.
Current source IOUT is the output load current. The driver circuit uses the input
signal on-time width to generate two control signals VUG and VLG, these two signals
should be avoid to turn on at the same time, because this operation make this
system to have a shoot through problem. The compensator of RC and CC should be
designed an optimization to increase the transient response. Only the feedback
signal VFB and reference voltage VREF are built inside the IC. The rising slope of
inductor current IL is sensed and required in the feedback loop as it contains supply
information for the current-mode control circuit. The ramp generator circuit that can
generate voltage signal VRAMP to be added to inductor current IL is sampled. The
sum voltage of voltage signal VRAMP and voltage signal VCS is used through a
resistor to obtain voltage signal VSUM. The output signal of the comparator depends
on the results of input signals VCOMP and VSUM.
The PWM three-terminal model [23, 24] with the current-mode control circuit for
the boost converter is shown in Fig. 2.15. In Fig. 2.15, a transfer function for
control-to-output GCTO(s) can be obtained by Eqs. (2.29)–(2.45). FP(s) provides
one pole, one zero, and one RHP zero by Eqs. (2.34)–(2.38) [14–22]. The pole of
FP(s) is located at the dominant low-frequency characteristic of the system by
Eqs. (2.34)–(2.35). FH(s) provides two poles at half the switching frequency with a
good ramp compensation design by Eqs. (2.39)–(2.41). The high-frequency effects
52 2 Review of the PWM Control Circuits for Power Converters

Fig. 2.15 PWM L


three-terminal model with c
current-mode control circuit
D
for boost converter iL i OUT
p R
CO
1
-VOUT ROUT V
d OUT

D I Ld CO

d
Fm

Kr VOUT

iL
HE(s) Ri
FC VCOMP

are ignored. However, if the mC value is small, it presents no ramp compensation or


small ramp compensation. The current-mode control is more stable than the voltage-
mode control with only one dominant pole.

V^OUT ð1  DÞ  ROUT
GCTO ðsÞ ¼ ¼  FpðsÞ  FhðsÞ ð2:29Þ
^
VCOMP KD  Ri

ð1  DÞ  ROUT
GCTO ð0Þ ¼ ð2:30Þ
KD  Ri
 
ð1  DÞ2  ROUT 1 K
KD ¼ 2 þ þ ð2:31Þ
KD  Ri Km ð1  DÞ

1
Km ¼ ð2:32Þ
ð0:5  DÞ  Ri  TLS þ VRAMP
VOUT

TS
K ¼ 0:5  Ri   ð1  DÞ  D ð2:33Þ
L
2.2 Current-Mode Control Circuit for Power Converters 53

ð1 þ s=SZ1 Þ  ð1  s=SZ2 Þ
FpðsÞ ¼ ð2:34Þ
1 þ s=SP

KD
SP ¼ ð2:35Þ
CO  ROUT

1
SZ1 ¼ ð2:36Þ
CO  RCO

ROUT
SZ2 ¼ ð2:37Þ
LD
L
LD ¼ ð2:38Þ
ð1  D Þ2

1
FhðsÞ ¼ ð2:39Þ
1 þ xn QP þ xs 2
s 2

1
QP ¼ h  i ð2:40Þ
p  1þ SE
SN  D0  0:5

xn ¼ p  FS ð2:41Þ

FC ¼ 1 ð2:42Þ

SE
mC ¼ 1 þ ð2:43Þ
SN

VIN
SN ¼  Ri ð2:44Þ
L
SE ¼ VRAMP  FS ð2:45Þ

The first zero SZ1 depends on output capacitor CO and ESR RCO of the output
capacitor. Low ESR of a ceramic capacitor for output capacitors is less than 10 mX,
and the first zero SZ1 location is set at a high frequency of the current-mode control
circuit for the boost converter. Thus, the control-to-output GCTO(s) is changed to
one pole and one RHP zero system. The second zero SZ2 is an RHP zero. RHP zero
has the same 20 dB/decade rising gain magnitude as a conventional zero but with a
90° phase lag instead of lead. This characteristic is difficult, if not impossible, to
compensate. The designer is usually forced to roll off the loop gain at a relatively
low frequency. The crossover frequency may be a decade or more below what it
otherwise could be, resulting in severe impairment of the dynamic response. In
general, RHP zero is at a higher frequency than the dominant pole for the boost
54 2 Review of the PWM Control Circuits for Power Converters

Table 2.4 Operation conditions in voltage-mode control circuit for boost converter
VIN (V) 5V FS (kHz) 1200 kHz L (lH) 470 nH
VOUT (V) 13.6 V RD1 (kX) 26.9 kX CO (lF) 80 lF
IOUT (A) 0.6 A RD2 (kX) 2.7 kX RCO (mX) 2 mX
VREF (V) 1.24 V Ri (A/V) 0.1 A/V VRAMP (V) 0.872 mV

converter. The RHP zero phase drop starts a decade earlier and negatively affects
the potential phase margin of the converter’s control loop.
Table 2.4 lists the operation conditions of the current-mode control circuit for
the boost converter. Based on Table 2.4, a comparison of MathCAD predictions
and SIMPLIS simulation results of the open-loop control-to-output bode plot for the
boost converter is shown in Fig. 2.16. In Fig. 2.16, the MathCAD predictions
verify the SIMPLIS simulation results. The red-colored line represents the

(a) Gain

(b) Phase

Fig. 2.16 Comparison of MathCAD predictions and SIMPLIS simulation results open-loop
control-to-output bode plot for boost converter
2.2 Current-Mode Control Circuit for Power Converters 55

MathCAD predictions, and the blue-colored dot represents the SIMPLIS simulation
results. The DC gain curve of the open-loop control-to-output bode plot is equal to
24.386 dB by Eq. (2.30) using MathCAD, and the DC gain curves of the open-loop
control-to-output bode plot using the SIMPLIS simulation tool is 23.38 dB. The
MathCAD predictions verify the SIMPLIS simulation results. Pole SP of
control-to-output is located at 442.06 Hz by Eq. (2.32). Based on this operation
condition using the 80 lF output capacitor, the first zero SZ1 is located at 994.7 kHz
by Eq. (2.33), and the second zero SZ2 is located at 1010 kHz by Eq. (2.34). At this
operation condition, the open-loop control-to-output bode plot shows one pole at
the dominant low-frequency characteristics of the system only, so the current-mode
control is more stable than the voltage-mode control with only one dominant pole.
RHP zero SZ2 depends on the output loading and duty cycle, so the user needs to
consider an optimal compensator at the worst conditions to ensure that RHP zero
SZ2 does not affect system loop stability.

2.3 Compensation Design for Power Converters

The voltage-mode and current-mode control circuits for power converters can
obtain different control-to-output transfer functions. The control-to-output transfer
function can indicate the number of poles and zeroes and their locations. In a power
converter, the system control loop is a closed-loop loop gain and has a feedback
loop to regulate the suitable duty cycle to control the power switches and obtain the
required output voltage and output loading. In general, feedback loop consists of
feedback resistors and a compensator. The purpose of the feedback resistors is to
design the required output voltage. The required output voltage depends on refer-
ence voltage VREF by Eq. (2.46).
 
RD1
VOUT ¼ VREF  1 þ ð2:46Þ
RD2

The purpose of adding a compensator to the error amplifier is to compensate for


some of the gains and phases contained in the control-to-output transfer function
and feedback resistors to achieve high stability of the power supply. The ultimate
goal is to make the overall closed-loop loop gain transfer function satisfy the
stability criteria. Figure 2.17 shows a block diagram of closed-loop loop gain for
the current-mode control circuit. Closed-loop loop gain T(s) contains the open-loop
control-to-output transfer function GCTO(s), the feedback resistors’ transfer function
GFB(s), and the compensator’s transfer function GCOMP(s).
In general, open-loop control-to-output transfer function GCTO(s) and the feed-
back resistors’ transfer function GFB(s) are fixed under operation conditions, so the
user or designer design the optimal compensator transfer function GCOMP(s)
according to the open-loop control-to-output transfer function GCTO(s) and the
feedback resistors’ transfer function GFB(s) to satisfy the stability criteria in
56 2 Review of the PWM Control Circuits for Power Converters

Fig. 2.17 Block diagram of VIN VOUT


closed-loop loop gain for Power Stage
current-mode control circuit VUG VLG Loop Gain T(s) IL
Ramp Current Feedback
Generator VRAMP VCS Sense Resistors

VSUM VFB
Comparator
Compensator
VCOMP

Eq. (2.47). Based on this equation, the information of gain margin (GM) and phase
margin (PM) can be easily obtained to understand system loop stability.

1 þ T ðsÞ\0 ð2:47Þ

This stability criterion of gain margin is the phase of closed-loop loop gain is
close to −180°, and the desired gain margin anywhere the gain is smaller than 0 dB
by Eq. (2.48). The gain margin is not defined and considered a safe value. In
general, the gain margin is smaller than −10 dB at the least, as shown in Fig. 2.18.
If the phase of closed-loop loop gain is not smaller than −180°, the gain margin
would be difficult to find.

GM ¼ 0 dB  T ðsÞj\TðsÞ¼180 ð2:48Þ

The stability criterion of the gain margin of closed-loop loop gain is close to
0 dB, and the desired phase margin anywhere the phase is greater than −180° by

Fig. 2.18 Definition of Gain Gain


margin and phase margin for (dB)
closed-loop loop gain

BW Frequency
0 dB (Hz)
Gain Margin
Phase
(Degrees)

Frequency
-90O (Hz)

Phase Margin
-180O
2.3 Compensation Design for Power Converters 57

Eq. (2.49). The phase margin is generally considered a safe value when it is greater
than 45° and yields a well-damped transient load response, as shown in Fig. 2.18.

PM ¼ 180 0 dB þ \Tðs)jjTðsÞj¼0 dB ð2:49Þ

Having the slope of the gain curve at crossover frequency FC with a value of
−20 dB/decade is desirable because the phase variation is not a sharp phase drop of
180° at the crossover frequency, and it exerts an effect on reducing the phase
margin. The slope of the gain curve is designed to be −40 dB/decade to exceed
crossover frequency FC and to increase noise immunity. The impact on the phase
margin must be avoided because a good phase margin is prioritized over good noise
immunity.
An optimal compensator can allow this system to achieve high system loop
stability. It not only makes this system acquire good phase and gain margins but
also increases the DC gain of the closed-loop loop gain. In general,
control-to-output transfer function GCTO(s) and feedback resistors’ transfer function
GFB(s) possess a low DC gain. A low DC gain can cause steady-state errors, and
large steady-state errors exhibit poor load regulation, resulting in load regulation
results that are out of specifications for IC products. Moreover, large steady-state
errors may regulate a failure output voltage, thus causing the system to operate
under error conditions.
Figure 2.19 shows a schematic of the operational amplifier (OPA) type-II
compensator for AC analysis. The operational amplifier is the basis of the closed-
loop system and can be modeled similar to a voltage-controlled voltage source
(VCVS). In a feedback system, OPA amplifies the error detected between feedback
voltage VFB and reference voltage VREF. Under steady DC conditions, the input
terminals of feedback voltage VFB and reference voltage VREF are virtually at the
same voltage level. However, although both feedback resistors of the voltage

Fig. 2.19 Schematic of the LAC


OPA type-II compensator for
AC analysis
CC2

RC CC1

RD1 VX
VY
OPA
CAC
VREF
VAC
58 2 Review of the PWM Control Circuits for Power Converters

divider affect the DC level of the output voltage, from the AC point of view, only
feedback resistor RD1 is used in the AC analysis. Feedback resistor RD2 is con-
sidered and designed for output voltage requirement. Therefore, feedback resistor
RD2 is usually ignored in AC analysis. The OPA type-II compensator bode plot
cannot be directly simulated by SIMPLIS or HSpice without initial conditions, so
the simulation circuit usually uses inductor LAC and capacitor CAC to generate the
initial conditions. Then, the system’s bode plot can be simulated. Inductor LAC is
used to short output terminal VY to input terminal VX at the DC level, and the initial
conditions of voltage VY is equal to those of voltage VX. Moreover, capacitor CAC is
used to block the AC signal injected into input terminal VX. To avoid changes in the
OPA type-II compensator bode plot, inductor LAC should be designed with large
inductance of 1 GH, and capacitor CAC should be designed with large inductance of
1 GF.
In Fig. 2.19, a transfer function for OPA type-II compensator GCOMP(s) can be
obtained by Eqs. (2.50)–(2.53). Two poles and one zero are used for AC analysis,
as shown in Fig. 2.20. One pole is an initial pole, and this pole crosses over 0 dB at
SC. Regarding the location of pole SP1 and zero SZ1, zero SZ1 should be smaller than
pole SP1. The OPA type-II compensator may increase system loop stability, so
capacitor CC1 should be much larger than capacitor CC2 to ensure optimal system
loop stability. Moreover, pole SP1 is designed at a high frequency to decay the
high-frequency noise, so pole SP1 can increase the system loop to achieve good
noise immunity. In Fig. 2.20, the OPA type-II compensator has one zero to increase
the phase degrees, and the maximum phase degrees reach 90°. Thus, the OPA
type-II compensator is widely used as a voltage-mode control circuit for the buck
converter, current-mode control circuit for the buck converter, or current-mode
control circuit for the boost converter.

Fig. 2.20 Pole-Zero location Gain


of OPA type-II compensator (dB)
for AC analysis

0 dB rad/s
SZ1 SP1
Phase
(Degrees)

-90O rad/s
SZ1 SP1
2.3 Compensation Design for Power Converters 59

^Y
V ð1 þ s=SZ1 Þ
GCOMP ðsÞ ¼ ¼ ð2:50Þ
^X s  ðCC1  RD1 Þ  ð1 þ s=SP1 Þ
V

1
SP1 ¼ ð2:51Þ
CC2  RC

1
SZ1 ¼ ð2:52Þ
CC1  RC

1
SC ¼ ð2:53Þ
CC1  RD1

In power ICs, most products use an operational transconductance amplifier


(OTA) instead of an operational amplifier because the output signal of OTA is a
current signal and not a voltage signal. Hence, OTA does not need the circuit to
convert a current signal to a voltage signal. OTA can be modeled similar to a
VCCS. Figure 2.21 shows a schematic of OTA type-II compensator for AC anal-
ysis. In Fig. 2.21, resistor RC and capacitors CC1 and CC2 are not connected
between the input and output terminals. They are connected from the output ter-
minal to the ground. Resistor RO is an equivalent resistor at the output terminal and
is a factor that affects the DC gain of the OTA type-II compensator. OTA is the
basis of the closed-loop system. In a feedback system, OTA is used to amplify the
error detected between feedback voltage VFB and reference voltage VREF. Under
steady DC conditions, the input terminals of feedback voltage VFB and reference
voltage VREF are virtually at the same voltage level. However, although both
feedback resistors of the voltage divider affect the DC level of the output voltage,
from the AC point of view, feedback resistors RD1 and RD2 do not affect the AC
analysis. Therefore, feedback resistors RD1 and RD2 are usually ignored in AC
analysis.
In general, the OTA type-II compensator bode plot cannot be directly simulated
by SIMPLIS or HSpice without initial conditions. The simulation circuit usually
uses inductor LAC and capacitor CAC to generate the initial conditions. Then, the
bode plot of the system can be simulated. In Fig. 2.21, a transfer function for OTA

Fig. 2.21 Schematic of the LAC


OTA type-II compensator for
AC analysis
VX
VY
OTA
CAC
RC
CC2 RO VREF
VAC
CC1
60 2 Review of the PWM Control Circuits for Power Converters

Fig. 2.22 Pole-Zero location Gain


of OTA type-II compensator (dB)
for AC analysis

0 dB rad/s
SP0 SZ1 SP1
Phase
(Degrees)

-90O rad/s
SP0 SZ1 SP1

type-II compensator GCOMP(s) can be obtained by Eqs. (2.54)–(2.58). Two poles


and one zero are used for AC analysis, as shown in Fig. 2.22. The DC gain of the
OTA type-II compensator depends on gain GM of OTA and output equivalent
resistor RO, so the DC gain of the OTA type-II compensator is different from the
DC gain of the OPA type-II compensator. Pole SP0 is an initial pole located at a low
frequency.
Regarding the location of pole SP1 and zero SZ1, zero SZ1 should be designed
smaller than pole SP1. The OTA type-II compensator may increase system loop
stability, so capacitor CC1 should be much larger than capacitor CC2 to ensure
optimal system loop stability. Pole SP1 is designed at a high frequency to decay the
high-frequency noise, so pole SP1 can increase the system loop to achieve good
noise immunity. In Fig. 2.22, the OTA type-II compensator has one zero to increase
the phase degrees, and the maximum phase degrees reach 90O. Thus, the OTA
type-II compensator is widely used as a voltage-mode control circuit for the buck
converter, current-mode control circuit for the buck converter, or current-mode
control circuit for the boost converter.

^Y
V ð1 þ s=SZ1 Þ
GCOMP ðsÞ ¼ ¼ GM  RO ð2:54Þ
^X
V ð1 þ s=SP0 Þ  ð1 þ s=SP1 Þ

^Y
V
GCOMP ð0Þ ¼ ¼ GM  RO ð2:55Þ
^X
V

1
SP0 ¼ ð2:56Þ
CC1  RO
2.3 Compensation Design for Power Converters 61

1
SP1 ¼ ð2:57Þ
CC2  RC

1
SZ1 ¼ ð2:58Þ
CC1  RC

The OPA type-II compensator and OTA type-II compensator for AC analysis
have two poles and one zero structure. If the open-loop control-to-output has a
two-pole system at the dominant frequency, the compensator should be imple-
mented by the OPA type-III compensator or the OTA type-III compensator because
these compensators have three poles and two zeros structure. The designer or user
can design one pole at a high frequency or remove one pole to affect the closed-loop
loop gain.
Figure 2.23 shows a schematic of the OPA type-III compensator for AC anal-
ysis. The operational amplifier is the basis of the closed-loop system. Feedback
resistors RD1 and RD2 of the voltage divider affect the DC level of the output
voltage, but from the AC point of view, only feedback resistor RD1 is used in the
AC analysis. Feedback resistor RD2 is considered and designed for output voltage
requirement. Therefore, feedback resistor RD2 is usually ignored in AC analysis. In
general, the OPA type-III compensator bode plot cannot be directly simulated by
SIMPLIS or HSpice without initial conditions, so the simulation circuit usually uses
inductor LAC and capacitor CAC to generate the initial conditions. Then, the sys-
tem’s bode plot can be simulated.
In Fig. 2.23, a transfer function for OPA type-III compensator GCOMP(s) can be
obtained by Eqs. (2.59)–(2.64). Three poles and two zeros are used for AC analysis,
as shown in Fig. 2.24. One pole is an initial pole, and this pole crosses over 0 dB at
SC. Regarding the location of pole SP1 and zero SZ1, zero SZ1 should be designed
smaller than pole SP1. The OPA type-III compensator may increase the system loop
stability, so capacitor CC1 should be much larger than capacitor CC2 to ensure

Fig. 2.23 Schematic of the LAC


OPA type-III compensator for
AC analysis
CC2

RC CC1 RC3 CC3

RD1 VX
VY
OPA
CAC
VREF
VAC
62 2 Review of the PWM Control Circuits for Power Converters

Fig. 2.24 Pole-Zero location Gain


of OPA type-III compensator (dB)
for AC analysis

0 dB rad/s
SZ1 SZ2 SP1 SP2
Phase
(Degrees)

-90O rad/s
SZ1 SZ2 SP1 SP2

optimal system loop stability. Moreover, pole SP1 is designed at a high frequency to
decay high-frequency noise, so pole SP1 can increase the system loop to achieve
good noise immunity. In addition, regarding the location of pole SP2 and zero SZ2,
zero SZ2 must be formed smaller than pole SP2 by Eqs. (2.62)–(2.63) because the
sum of two resistors RC3 and RD1 are larger than resistor RC3 only.

^Y
V ð1 þ s=SZ1 Þ  ð1 þ s=SZ2 Þ
GCOMP ðsÞ ¼ ¼ ð2:59Þ
^
VX s  ðCC1  RD1 Þ  ð1 þ s=SP1 Þ  ð1 þ s=SP2 Þ

1
SP1 ¼ ð2:60Þ
CC2  RC

1
SZ1 ¼ ð2:61Þ
CC1  RC

1
SP2 ¼ ð2:62Þ
CC3  RC3

1
SZ2 ¼ ð2:63Þ
CC3  ðRC3 þ RD1 Þ

1
SC ¼ ð2:64Þ
CC1  RD1

In Fig. 2.24, the OPA type-III compensator has two zeros to increase the phase
degrees, and the maximum phase degrees reach 180°. Thus, the OPA type-III
2.3 Compensation Design for Power Converters 63

compensator is widely used as a voltage-mode control circuit for the buck converter
or voltage-mode control circuit for the boost converter. If open-loop control-
to-output has two poles at the dominant frequency, designers or users must apply
the OPA type-III compensator, and the closed-loop loop gain may suffer from a bad
phase margin with a large bandwidth or bad noise immunity. This implemented
circuit also needs two passive components for RC3 and CC3.
To increase the phase degrees in several real case designs for the OPA type-III
compensator, the designer or user usually removes pole SP2 of transfer function
GCOMP(s) for the OPA type-III compensator. A good option is to design resistor RC3
for an mX level or remove resistor RC3 directly. Two poles and two zeros are used
for AC analysis, as shown in Fig. 2.25. Transfer function GCOMP(s) can be obtained
and rewritten for the OPA type-III compensator by Eqs. (2.65)–(2.69).

^Y
V ð1 þ s=SZ1 Þ  ð1 þ s=SZ2 Þ
GCOMP ðsÞ ¼ ¼ ð2:65Þ
^
VX s  ðCC1  RD1 Þ  ð1 þ s=SP1 Þ

1
SP1 ¼ ð2:66Þ
CC2  RC

1
SZ1 ¼ ð2:67Þ
CC1  RC

Fig. 2.25 Pole-Zero location Gain


of OPA type-III compensator (dB)
without the resistor RC3 for
AC analysis

0 dB rad/s
SZ1 SZ2 SP1
Phase
(Degrees)

-90O rad/s
SZ1 SZ2 SP1
64 2 Review of the PWM Control Circuits for Power Converters

1
SZ2 ¼ ð2:68Þ
CC3  RD1

1
SC ¼ ð2:69Þ
CC1  RD1

As shown in Fig. 2.25, the OPA type-III compensator has two zeros to increase
the phase degrees, and the maximum phase degrees reach 180°. Thus, the OPA
type-III compensator is widely used as a voltage-mode control circuit for the buck
converter or voltage-mode control circuit for the boost converter. However, in the
OPA type-III compensator without resistor RC3, the closed-loop loop gain is needed
to check the performance in phase margin and noise immunity.
Figure 2.26 shows a schematic of the OTA type-III compensator for AC anal-
ysis. The operational amplifier is the basis of the closed-loop system. Feedback
resistors RD1 and RD2 of the voltage divider affect the DC level of the output
voltage, but from the AC point of view, only feedback resistor RD1 is used in the
AC analysis. It is different from the OTA type-II compensator for AC analysis.
Therefore, feedback resistor RD1 cannot be ignored in AC analysis. Feedback
resistor RD2 is considered and designed for output voltage requirement, so feedback
resistor RD2 is usually ignored in AC analysis. In general, the OTA type-III com-
pensator bode plot cannot be directly simulated by SIMPLIS or HSpice without
initial conditions, so the simulation circuit usually uses inductor LAC and capacitor
CAC to generate the initial conditions. Then, the system’s bode plot can be
simulated.
From Fig. 2.26, a transfer function for OTA type-III compensator GCOMP(s) can
be obtained by Eqs. (2.70)–(2.73). Three poles and two zeros are used for AC
analysis, as shown in Fig. 2.27. The DC gain of the OTA type-III compensator
depends on gain GM of OTA, output equivalent resistor RO, and feedback resistor
gain by Eq. (2.71). Thus, the DC gain of the OTA type-III compensator is different
from the DC gain of the OTA type-II compensator. Pole SP0 is an initial pole, and
this pole is located at a low frequency.

Fig. 2.26 Schematic of the LAC


OTA type-III compensator for
AC analysis RC3 CC3

RD1 VX
VY
OTA
CAC
RC RD2
CC2 RO VREF
VAC
CC1
2.3 Compensation Design for Power Converters 65

Fig. 2.27 Pole-Zero location Gain


of OTA type-III compensator (dB)
for AC analysis

0 dB rad/s
SP0 SZ1 SZ2 SP1 SP2
Phase
(Degrees)

-90O rad/s
SP0 SZ1 SZ2 SP1 SP2

Regarding the location of pole SP1 and zero SZ1, zero SZ1 should be designed
smaller than pole SP1. The OTA type-III compensator may increase the system loop
stability, so capacitor CC1 should be designed much larger than capacitor CC2 to
ensure optimal system loop stability. Moreover, pole SP1 is designed at a high
frequency to decay the high-frequency noise, so pole SP1 can increase the system
loop to achieve good noise immunity. Regarding the location of pole SP2 and zero
SZ2, zero SZ2 must be formed smaller than pole SP2 by Eqs. (2.75)–(2.76). Several
real case designs for the OTA type-III compensator are used to simply design the
location of pole SP2 and zero SZ2. In general, assuming that feedback resistors RD1
and RD2 are much larger than resistor RC3, the location of pole SP2 and zero SZ2 for
transfer function GCOMP(s) can be obtained and rewritten for the OTA type-III
compensator by Eqs. (2.77)–(2.78). The location of pole SP2 and zero SZ2 depends
on capacitor CC3 and feedback resistors RD1 and RD2, so the location between pole
SP2 and zero SZ2 depends on the output voltage. In this case, capacitor CC3 can be
defined as a feedforward capacitor. A feedforward capacitor is widely used to
parallel feedback resistor RD1 and to increase the phase margin for the OPA type-II
compensator and OTA type-II compensator.
If the output voltage is high and the location between pole SP2 and zero SZ2 is
large, then better phase degrees can be obtained. Thus, if the output voltage is low
and the location between pole SP2 and zero SZ2 is small, zero SZ2 cannot contribute
to the phase degrees. For this reason, if the output voltage is low, using the OTA
type-III compensator instead of OTA type-II compensator is not good because extra
66 2 Review of the PWM Control Circuits for Power Converters

cost is needed for two passive components for RC3 and CC3 to achieve the same
performance in phase margin.

^Y
V RD2 ð1 þ s=SZ1 Þ  ð1 þ s=SZ2 Þ
GCOMP ðsÞ ¼ ¼ GM  RO  
^
VX RD1 þ RD2 ð1 þ s=SP0 Þ  ð1 þ s=SP1 Þ  ð1 þ s=SP2 Þ
ð2:70Þ

^Y
V RD2
GCOMP ð0Þ ¼ ¼ GM  RO  ð2:71Þ
^
VX RD1 þ RD2

1
SP0 ¼ ð2:72Þ
CC1  RO

1
SP1 ¼ ð2:73Þ
CC2  RC

1
SZ1 ¼ ð2:74Þ
CC1  RC

1
SP2 ¼ h i ð2:75Þ
ðRD1 RD2 þ RC3 RD2 þ RD1 RC3 Þ
CC3  ðRD1 þ RD2 Þ

1
SZ2 ¼ ð2:76Þ
CC3  ðRC3 þ RD1 Þ

1
SP2 ¼   ð2:77Þ
RD1 RD2
CC3  RD1 þ RD2

1
SZ2 ¼ ð2:78Þ
CC3  RD1

As shown in Fig. 2.27, the OTA type-III compensator has two zeros to increase
the phase degrees, and the maximum phase degrees reach 180°. Thus, the OTA
type-III compensator is widely used as a voltage-mode control circuit for the buck
converter or voltage-mode control circuit for the boost converter. If open-loop
control-to-output has two poles at the dominant frequency, designers or users may
apply the OTA Type-III compensator, and the closed-loop loop gain may suffer
from a bad phase margin with a large bandwidth or bad noise immunity. Pole SP2
cannot be removed from transfer function GCOMP(s) for the OTA type-III com-
pensator because pole SP2 does not depend on resistor RC3 only; it depends on
feedback resistors RD1 and RD2, so the OTA type-III compensator cannot achieve
two poles and two zeros, similar to the OPA type-III compensator.
2.4 Experimental Results 67

2.4 Experimental Results

MathCAD predictions, SIMPLIS simulation results, and experimental verifications


were conducted to determine the feasibility and performance of the proposed the
current-mode control circuit with OTA Type-II compensator for buck converter.
The specifications are as follows:
(1) Input DC voltage range (VIN): 12 V
(2) Output DC voltage range (VOUT): 3.3 V
(3) Output load current (IOUT): 3A
(4) Switching frequency (FS): 350 kHz
(5) Feedback resistors (RD1, RD2): 25.7 and 10 kX
(6) Main inductor (L): 10 lH
(7) Output capacitors (CO): 22 lF/25 V * 2 (RCO: 10 mX)
(8) Reference voltage (VREF): 0.925 V
(9) Fixed ramp voltage (VRAMP): 0.507 V
(10) Current Sense Ratio (Ri): 0.2 A/V
(11) OTA Type-II compensator circuit: GM ¼ 1:25 mV/A; RO ¼ 200 MX;
RC ¼ 5:9 kX; CC1 ¼ 6:2 nF; CC2 ¼ 158 pF.
Figure 2.28 shows the circuit structure of the current-mode control circuit with
the OTA type-II compensator for the buck converter. Control-to-output transfer
function GCTO(s) for the current-mode control circuit of the buck converter has three
poles and one zero. FP(s) has one pole and one zero. The pole of FP(s) is located at
the dominant low-frequency characteristics of the system by Eqs. (2.19)–(2.20).
MathCAD predictions of the bode plots of control-to-output, feedback resistor,
and compensator are shown in Fig. 2.29. The red-colored curve represents the OTA
type-II compensator bode plot, the blue-colored curve represents the control-to-
output bode plot, and the brown-colored curve represents the feedback resistors’
bode plot.

Fig. 2.28 Circuit structure of S1 VLX L VOUT


current-mode control circuit
with OTA type-II VIN VUG IL
VLG S2 RCO IOUT
compensator for buck
converter Ri CO
RD1
VCS
Driver VRAMP Ramp
Generator
VSUM
VCOMP VFB
OTA

RC VREF RD2
CC2
CC1
68 2 Review of the PWM Control Circuits for Power Converters

Gain (dB)

Frequency (Hz)
(a) Gain
Phase (Degrees)

Frequency (Hz)
(b) Phase

Fig. 2.29 MathCAD predictions of the bode plots of control-to-output, feedback resistor, and
compensator

Based on the specifications, pole SP is located at 4.3 kHz by Eq. (2.20), and zero
SZ1 is located at 723 kHz by Eq. (2.21). Zero SZ1 depends on output capacitor CO
and ESR RCO of the output capacitor, so zero SZ1 cannot contribute to the phase
degrees because the output capacitor uses a low ESR RCO of the ceramic capacitor.
2.4 Experimental Results 69

The transfer function for OTA type-II compensator GCOMP(s) obtained by


Eqs. (2.54)–(2.58) gives two poles and one zero. The DC gain of the OTA type-II
compensator depends on gain GM of OTA and output equivalent resistor RO. Based
on the specifications, the DC gain of the OTA type-II compensator is 108 dB by
Eq. (2.55). The DC gain of the OTA type-II compensator can be designed to
achieve optimal bandwidth. In general, the optimal bandwidth is equal to a range
between 0.1 * FS and 0.15 * FS when switching frequency FS is smaller than
1 MHz. If switching frequency FS is larger than or equal to 1 MHz and the optimal
bandwidth is close to 0.1 * FS. A low bandwidth can obtain a slow load transient
response, and a large bandwidth can result in poor noise immunity or unstable
control loop. In addition, the DC gain of the OTA type-II compensator can be
designed in consideration of control-to-output transfer function GCTO(s) and feed-
back resistor transfer function GFB(s).
With regard to OTA type-II compensator GCOMP(s), pole SP0 is an initial pole
and located at a low frequency. Based on the specifications, pole SP0 of the OTA
type-II compensator is located at 0.127 Hz by Eq. (2.56). The location of pole SP0
depends on the output equivalent resistor RO, and pole SP0 can be similar to an
initial pole. Regarding the location of pole SP1 and zero SZ1, zero SZ1 should be
designed smaller than pole SP1. The OTA type-II compensator may increase the
system loop stability, so capacitor CC1 should be much larger than capacitor CC2 to
ensure optimal system loop stability. To cancel pole SP of the control-to-output
transfer function GCTO(s), zero SZ1 of the OTA type-II compensator can be
designed with the same location as pole SP at 4.3 kHz. Moreover, pole SP1 can be
designed at a high frequency to decay the high-frequency noise, so pole SP1 can
increase the system loop to achieve good noise immunity. Pole SP1 is located at a
half of switching frequency to ensure that pole SP1 does not affect the closed-loop
loop gain.
Regarding the feedback resistors’ transfer function GFB(s) by Eq. (2.79), this
equation is without any pole and zero, so GFB(s) is a DC gain with 0 phase degree.
Based on the specifications, the DC gain of the feedback resistors is −11.05 dB.

VREF RD2
GFB ðsÞ ¼ ¼ ð2:79Þ
VOUT RD1 þ RD2

Figure 2.30 shows the perturbation injection circuit at the output voltage ter-
minal to measure the bode plot of closed-loop loop gain. If feedback resistors RD1
and RD2 are designed into the integrated circuit, closed-loop loop gain cannot be
measured by a network or spectrum analyzer. However, if feedback resistors RD1
and RD2 are placed on the evaluation board to determine the output voltage,
closed-loop loop gain can be measured by a network or spectrum analyzer. An AC
perturbation signal VAC as the input signal can be generated by a network or
spectrum analyzer. VAC is a sinusoidal waveform, and the frequency of the sinu-
soidal waveform depends on the measurement frequency. In general, the mea-
surement frequency by a network or spectrum analyzer is between 1 kHz and
1 MHz.
70 2 Review of the PWM Control Circuits for Power Converters

Fig. 2.30 Perturbation S1 L VOUT


Perturbation
VLX Injection Circuit
injection circuit at the output
VIN VUG IL VA
voltage terminal to measure VLG RCO
S2 IOUT VAC
the bode plot of closed-loop Ri CO 50Ω
loop gain
VCS
Driver VB
VRAMP Ramp
Generator RD1
VSUM
VCOMP VFB
OTA

RC VREF RD2
CC2
CC1

AC perturbation signal VAC can be injected by a transformer, and the bandwidth


of the transformer should cover a range between 1 kHz and 1 MHz to obtain a
correct closed-loop loop gain bode plot. To achieve impedance matching, the
output terminal should be placed at 50 X parallel to the transformer. The two output
terminals from the network or spectrum analyzer should be connected to terminals
VA and VB individually.
Figure 2.31 compares the MathCAD predictions, simulation results, and
experimental results for the closed-loop loop gain bode plot of the current-mode
control circuit with the OTA type-II compensator for the buck converter. The
red-colored solid line represents the MathCAD predictions for the closed-loop loop
gain bode plot, the blue-colored dotted line represents the SIMPLIS simulation
results for the closed-loop loop gain bode plot, and the green-colored dashed line
represents the measurement results for the closed-loop loop gain bode plot. The
MathCAD predictions and SIMPLIS simulation results are similar to the experi-
mental results.
The MathCAD predictions, SIMPLIS simulation results, and experimental
results confirm that the current-mode control circuit with the OTA type-II com-
pensator for the buck converter can significantly avoid the subharmonic issue. The
bandwidth of closed-loop loop gain is equal to 0.1 * FS, and the bandwidth is
35 kHz. The phase margin is 50°, and the system control loop is stable. The gain
margin cannot be measured in Fig. 2.31.
To further examine the advantages and superiority of the LED driver for the
boost converter. LED applications for color and white high-brightness LEDs are
expanding into markets, similar to smartphones to LCD TVs, architectural lighting,
and general lighting. LEDs are connected in series in the output terminal, which
requires a higher driving voltage level. However, most portable devices are pow-
ered by single lithium-ion battery packs, with voltage levels ranging from 2.5 to
4.2 V. Therefore, an LED driver in a portable device must step up the voltage level
to drive the high-voltage LED string. Thus, a portable device needs a boost con-
verter, not a buck converter, to drive the LED.
2.4 Experimental Results 71

80

60

40

20
Gain (dB)

− 20

− 40 Predicted Curve
Simulated Curve
− 60
Measured Curve
− 80
3 4 5 6
100 1 ×10 1× 10 1 ×10 1× 10
Frequency (Hz)
(a) Gain

135

90
Phase (Degrees)

45

Predicted Curve
− 45
Simulated Curve
Measured Curve
− 90
3 4 5 6
100 1×10 1 ×10 1 ×10 1×10
Frequency (Hz)
(b) Phase

Fig. 2.31 MathCAD predictions, simulation results, and experimental results for the closed-loop
loop gain bode plot of current-mode control circuit with OTA type-II compensator for buck
converter
72 2 Review of the PWM Control Circuits for Power Converters

MathCAD predictions, SIMPLIS simulation results, and experimental verifica-


tions were conducted to determine the feasibility and performance of the proposed
LED driver of the current-mode control circuit with OTA Type-II compensator for
boost converter. The specifications are as follows:
(1) Input DC voltage range (VIN): 3.7 V
(2) Output DC voltage range (VOUT): 34 V (10 WLEDs)
(3) Output load current (IOUT): 20 mA
(4) Switching frequency (FS): 1 MHz
(5) Feedback resistor (RD2): 5.2 X
(6) Main inductor (L): 22 lH
(7) Output capacitors (CO): 1 lF/50 V * 1 (RCO: 10 mX)
(8) Reference voltage (VREF): 0.104 V
(9) Fixed ramp voltage (VRAMP): 0.75 V
(10) Current Sense Ratio (Ri): 0.526 A/V
(11) OTA Type-II compensator circuit: GM ¼ 1 mV/A; RO ¼ 270 MX;
RC ¼ 99 kX; CC1 ¼ 8:53 nF; CC2 ¼ 11 pF.
LED is similar to a diode in terms of behavior, so if the output loading is changed
from pure resistors to LED, and the control loop must be designed to regulate output
current IOUT and not output voltage VOUT. Based on the specifications, reference
voltage VREF is 0.104 V, so feedback voltage VFB should be equal to 0.104 V, as
shown in Fig. 2.32. Feedback resistor RD2 is designed to regulate output current
IOUT, which is equal to 20 mA, so feedback resistor RD2 is 5.2 X. Given that the
output loading is 10 WLEDs and the control loop regulates output current IOUT at
20 mA, the LEDs are under constant current, which is different from the scenario in a
pure resistor. Moreover, LEDs do not need variation current, so the LED driver of
the current-mode control circuit with the OTA type-II compensator for the boost
converter has no load transient response requirements. This is the reason the
current-mode control circuit with the OTA type-II compensator for the boost con-
verter for LED applications is without any loop stability concern.
Figure 2.32 shows the circuit structure LED driver of the current-mode control
circuit with OTA Type-II compensator for boost converter. The control-to-output
transfer function GCTO(s) for current-mode control circuit for boost converter gives
three poles and two zeros. FP(s) gives one pole, one zero, and one RHP zero by
Eqs. (2.34)–(2.38). The pole of FP(s) is located at the dominant low frequency
characteristics of the system by Eqs. (2.34)–(2.35). Based on one pole at the
dominant low frequency, the OTA Type-II compensator or the OPA Type-II
compensator is suitable to implement with current-mode control circuit for boost
converter.
MathCAD predictions of the bode plots of control-to-output, feedback resistor,
and compensator are shown in Fig. 2.33. The red-colored curve represents the OTA
type-II compensator bode plot, the blue-colored curve represents the control-to-
output bode plot, and the brown-colored curve represents the feedback resistors’
bode plot.
2.4 Experimental Results 73

Fig. 2.32 Circuit structure of L VLX S1 VOUT


VIN
LED driver of the
current-mode control circuit IL VUG IL
RCO
with OTA type-II VLG S2 IOUT
VUG Ri CO
compensator for boost
converter VCS 10
Driver VRAMP Ramp WLEDs
Generator
VSUM
VCOMP VFB
OTA

CC2
RC VREF RD2

CC1

Based on the specifications, the DC gain curves of the open-loop


control-to-output bode plot is equal to 41.8 dB by Eq. (2.30) using MathCAD.
Pole SP of control-to-output is located at 187.2 Hz by Eq. (2.32). Based on this
operation condition using the 1 lF output capacitor, the first zero SZ1 is located at
13 MHz by Eq. (2.33). The second zero SZ2 is located at 145.6 kHz by Eq. (2.34).
In this operation condition, the open-loop control-to-output bode plot shows one
pole at the dominant low-frequency characteristics of the system only, so the
current-mode control is more stable than the voltage-mode control with one
dominant pole only.
The transfer function for the OTA type-II compensator GCOMP(s) by
Eqs. (2.54)–(2.58) gives two poles and one zero. The DC gain of the OTA type-II
compensator depends on gain GM of OTA and output equivalent resistor RO. Based
on the specifications, the DC gain of the OTA type-II compensator is 108.63 dB by
Eq. (2.55). The DC gain of the OTA type-II compensator can be designed to
achieve optimal bandwidth with the RHP zero limitation. Based on the RHP zero
limitation, the optimal bandwidth may not reach a range between 0.1 * FS and
0.15 * FS when switching frequency FS is smaller than 1 MHz, so designing the
optimal bandwidth of the boost converter to be larger than that of the buck con-
verter is difficult. In addition, the DC gain of the OTA type-II compensator can be
designed considering control-to-output transfer function GCTO(s) and feedback
resistor transfer function GFB(s).
In OTA type-II compensator GCOMP(s), pole SP0 is an initial pole and located at a
low frequency. Based on the specifications, pole SP0 of the OTA type-II compen-
sator is located at 0.07 Hz by Eq. (2.56). The location of pole SP0 depends on
output equivalent resistor RO. Pole SP0 can be similar to an initial pole. Regarding
the location of pole SP1 and zero SZ1, zero SZ1 should be designed smaller than pole
74 2 Review of the PWM Control Circuits for Power Converters

Gain (dB)

Frequency (Hz)
(a) Gain
Phase (Degrees)

Frequency (Hz)
(b) Phase

Fig. 2.33 MathCAD predictions of the bode plots of control-to-output, feedback resistor, and
compensator

SP1. The OTA type-II compensator may increase the system loop stability, so
capacitor CC1 should be much larger than capacitor CC2 to ensure optimal system
loop stability. To cancel pole SP of the control-to-output transfer function GCTO(s),
zero SZ1 of the OTA type-II compensator can be designed with the same location as
2.4 Experimental Results 75

Perturbation
L VLX S1 VOUT
Injection Circuit
VIN
VA
IL VUG IL
RCO VAC
VLG S2 50
VUG Ri CO IOUT

VCS VB
Driver VRAMP 10
Ramp
WLEDs
Generator
VSUM
VCOMP VFB
OTA

CC2
RC VREF RD2

CC1

Fig. 2.34 Perturbation injection circuit at the output voltage terminal to measure the bode plot of
closed-loop loop gain

pole SP at 187.2 Hz. Moreover, pole SP1 is designed at a high frequency to decay
high-frequency noise, so pole SP1 can increase the system loop to achieve good
noise immunity. Pole SP1 is located at 145 kHz to ensure that pole SP1 does not
affect the closed-loop loop gain.
Regarding the feedback resistors’ transfer function GFB(s) by Eq. (2.79), this
equation is without any pole and zero, so GFB(s) is a DC gain with 0 phase degree.
Based on the specifications, the DC gain of the feedback resistors is −50.3 dB.
Figure 2.34 shows the perturbation injection circuit at the output voltage ter-
minal to measure the bode plot of closed-loop loop gain. The output loading of
LEDs should be placed on the evaluation board and feedback resistor RD2 to
determine the output current. Closed-loop loop gain can be measured by a network
or spectrum analyzer. AC perturbation signal VAC as the input signal can be gen-
erated by a network or spectrum analyzer. AC perturbation signal VAC is a sinu-
soidal waveform, and the frequency of this sinusoidal waveform depends on the
measurement frequency. In general, the measurement frequency by a network or
spectrum analyzer is between 1 kHz and 1 MHz.
AC perturbation signal VAC can be injected by a transformer, and the bandwidth
of the transformer should cover a range between 1 kHz and 1 MHz to obtain a
correct closed-loop loop gain bode plot. To achieve impedance matching, the
output terminal should be placed at 50 X parallel to the transformer. The two output
terminals from the network or spectrum analyzer should be connected to terminals
VA and VB individually.
If the perturbation injection circuit is inserted into at the output voltage terminal
with output loading of LEDs, a problem would arise because the perturbation
injection circuit should be placed at the high-impedance terminal. The output
76 2 Review of the PWM Control Circuits for Power Converters

L VLX S1 VOUT
VIN

IL VUG
IL RCO
VLG S2 IOUT
VUG CO
Ri 10
WLEDs
Driver VCS
VRAMP Ramp Perturbation
Generator Injection Circuit
VSUM VFB
VA
VCOMP
OTA VAC
50
VREF RD2
RC
CC2
VB
CC1

Fig. 2.35 Perturbation injection circuit at the input terminal of the OTA type-II compensator to
measure the bode plot of closed-loop loop gain

loading of LEDs is similar to the current source, so the output voltage terminal is a
low-impedance terminal, not a high-impedance terminal.
For this reason, the perturbation injection circuit should be placed at the input
terminal of the OTA type-II compensator because this input terminal is a
high-impedance terminal and a path of closed-loop loop gain. Thus, the perturba-
tion injection circuit at the input terminal of the OTA type-II compensator can
measure the correct bode plot of closed-loop loop gain, as shown in Fig. 2.35.
Figure 2.36 compares the MathCAD predictions, simulation results, and
experimental results for the closed-loop loop gain bode plot of the current-mode
control circuit with the OTA type-II compensator for the boost converter. The
red-colored solid line represents the measurement results for the closed-loop loop
gain bode plot, the blue-colored dotted line represents the MathCAD predictions for
the closed-loop loop gain bode plot, and the green-colored dashed line represents
the SIMPLIS simulation results for the closed-loop loop gain bode plot. The
MathCAD predictions and SIMPLIS simulation results are similar to the experi-
mental results. The MathCAD predictions, SIMPLIS simulation results, and
experimental results confirm that the current-mode control circuit with the OTA
type-II compensator for the boost converter can significantly prevent the subhar-
monic issue. In addition, the bandwidth of closed-loop loop gain hardly reaches
0.1 * FS, and the bandwidth is 65 kHz. The phase margin is 46°, and the system
control loop is stable. The gain margin cannot be measured in Fig. 2.36.
2.4 Experimental Results 77

(a) Gain

(b) Phase
Fig. 2.36 MathCAD predictions, simulation results, and experimental results for the closed-loop
loop gain bode plot of current-mode control circuit with OTA type-II compensator for boost
converter
78 2 Review of the PWM Control Circuits for Power Converters

2.5 Summary

Experimental results, SIMPLIS simulation results, and MathCAD predictions


confirm that closed-loop loop gain with the well compensation design of the
current-mode control circuit with OTA Type-II compensator for buck converter
does not only maintain system stability, but also significantly prevents subhar-
monic. Moreover, the closed-loop loop gain of the proposed LED driver of the
current-mode control circuit with OTA Type-II compensator for boost converter
can be verified by experimental results, SIMPLIS simulation results, and MathCAD
predictions to achieve the good loop stability and to prevent subharmonic with the
optimal compensation design.

References

1. C.J. Chen, D. Chen, C.W. Tseng, C.T. Tseng, Y.W. Chang, K.C. Wang, A novel ripple-based
constant on-time control with virtual inductor current ripple for buck converter with ceramic
output capacitors, in Proceedings of IEEE Applied Energy Conversion Congress and
Exposition Conference (2011), pp. 1244–1250
2. W.H. Ki, Signal flow graph in loop gain analysis of DC-DC PWM CCM switching
converters. IEEE Trans. Circuits Syst. Part I 45(6), 644–654 (1998)
3. W.H. Ki, Analysis of subharmonic oscillation of fixed-frequency current-programming switch
mode power converters. IEEE Trans. Circuits Syst. Part I 45(1), 104–108 (1998)
4. A.D. Schoenfeld, Y. Yu, ASDTIC control and standardized interface circuits applied to buck,
parallel and buck-boost DC-to-DC power converters, in NASA, Washington, DC, NASA
Rep. NASA CR-121106, Feb 1973
5. C.W. Deisch, Switching control method changes power converter into a current source, in
Proceedings of IEEE Power Electronics Specialists Conference (1978), pp. 300–306
6. P.L. Hunter, Converter circuit and method having fast responding current balance and
limiting, U.S. Patent 4 002 963, 1 Nov 1977
7. L.H. Dixon, Average current-mode control of switching power supplies, in Proceedings of
Unitrode Power Supply Design Seminar Handbook (1990), pp. 5.1–5.14
8. N. Mohan, Power electronics circuits: an overview, in Proceedings of IEEE Industrial
Electronics Society Conference (1988), pp. 522–527
9. N. Mohan, W.P. Robbins, P. Imbertson, T.M. Undeland, R.C. Panaitescu, A.K. Jain, P. Jose,
T. Begalke, Restructuring of first courses in power electronics and electric drives that
integrates digital control. IEEE Trans. Power Electron. 18, 429–437 (2003)
10. R.D. Middlebrook, S. Cuk, A general unified approach to modeling switching-converter
power states, in Proceedings of IEEE Power Electronics Specialists Conference (1976),
pp. 73–86
11. D.Y. Chen, H.A. Owen, T.G. Wilson, Computer Aided design and graphics applied to the
study of inductor-energy-storage DC-to-DC electronic power converters. IEEE Trans.
Aerosp. Electron. Syst. AES-9, 585 (1973)
12. P. Burger, Analysis of a class of pulse modulated DC-to-DC power converters. IEEE Trans.
Ind. Electron. Control Instrum. IECI-22, 104 (1975)
13. W.H. Lau, H. Chung, C.M. Wu, N.K. Poon, Realization of digital audio amplifier using
zero-voltage-switched PWM power converter. IEEE Trans. Circuits Syst. Part I 47(3), 303–
311 (2000)
References 79

14. R.B. Ridley, A new, continuous-time model for current-mode control. IEEE Trans. Power
Electron. 6, 271–280 (1991)
15. R.B. Ridley, A new continuous-time model for current-mode control with constant on-time,
constant off-time, and discontinuous conduction mode, in Proceedings of IEEE Power
Electronics Specialists Conference (1990), pp. 382–389
16. J. Li, F.C. Lee, New modeling approach and equivalent circuit representation for
current-mode control. IEEE Trans. Power Electron. 25, 1218–1230 (2010)
17. J. Li, Current-mode control: Modeling and its digital application, Ph.D. dissertation, Bradley
Department of Electrical and Computer Engineering, Virginia Polytechnic Institute State
University, Blacksburg, VA, USA, 2009
18. R.W. Erickson, D. Maksimovic, Fundamentals of Power Electronics (Kluwer, Norwell, MA,
2001)
19. W.W. Chen, J.F. Chen, T.J. Liang, L.C. Wei, W.Y. Ting, Designing a dynamic ramp with an
invariant inductor in current-mode control for an on-chip Buck converter. IEEE Trans. Power
Electron. 29(2), 750–758 (2014)
20. W.W. Chen, J.F. Chen, T.J. Liang, L.C. Wei, W.Y. Ting, Dynamic Ramp with the invariant
inductor in current-mode control for Buck converter, in Proceedings of IEEE APEC (2013),
pp. 1244–1249
21. W.W. Chen, J.F. Chen, T.J. Liang, Dynamic Ramp control in current-mode adaptive on-time
control for Buck converter on chip, in Proceedings of IEEE Future Energy Electronics
Conference and ECCE Asia (IFEEC 2017 - ECCE Asia) (2017), pp. 280–285
22. S. Robert, H. Anatole, Understanding and applying current-mode control theory, in Power
Electronics Technology Exhibition and Conference (2007)
23. V. Vorperian, Simplified analysis of PWM converters using model of PWM switch-I.
Continuous conduction mode. IEEE Trans. Aerosp. Electron. Syst. 26(3), 490–496 (1990)
24. V. Vorperian, Simplified analysis of PWM converters using model of PWM switch-II.
Discontinuous conduction mode. IEEE Trans. Aerosp. Electron. Syst. 26(3), 490–496 (1990)
25. F. Tian, S. Kasemsan, I. Batarseh, An adaptive slope compensation for the single-stage
inverter with peak current-mode control. IEEE Trans. Power Electron. 26(10), 2857–2862
(2011)
26. Z. Zansky, Current mode converter with controlled slope compensation, U.S. Patent 4 837
495, 6 June 1989
27. L. Yanming, L. Xinquan, C. Fuji, Y. Bing, J. Xinzhang, An adaptive slope compensation
circuit for buck DC-DC converter, in Proceedings of International Conference on ASIC, Oct
2007, pp. 608–611
Chapter 3
Designing a Dynamic Ramp with Invariant
Inductor in Current-Mode Control
Circuit for Buck Converter

3.1 Challenges for Wide Input Voltage Range

Switching DC–DC converters are widely used in the field of power conversion due
to their high efficiency and high power density. The popularity of smart phones,
tablet computer, notebook computer, and similar portable battery-powered elec-
tronic devices is increasing the demand for extended battery life, thereby driving
research toward high efficiency switching DC–DC converters.
In a DC–DC Buck converter, the current-mode control circuit is widely used to
achieve better transient response [1–13]. The current-mode control circuit contains
two feedback signals. The output voltage is fed into an error amplifier to generate a
control signal. The inductor current is sampled and compared with the control
signal to control the peak inductor current, thus regulating the output voltage to the
specified level. The inductor current feedback causes the control to have a better
transient response. However, it also produces a subharmonic issue when the duty
cycle is larger than 50%.
A ramp compensation signal is always added into the control loop to prevent
subharmonic [1–5, 11–16], as shown in Fig. 3.1. The ramp compensation is
designed according to the inductor current ripple determined by the input voltage,
output voltage, and inductance. A general control IC often imbeds a fixed ramp into
the control loop. However, this kind of chip is usually designed for a wide input and
output range, thus resulting in a varied inductor current ripple in the entire operation
range. Meeting the fixed ramp compensation requires that the inductance is changed
to maintain the same current ripple. Thus, the compensation circuit should also be
modified to keep system bandwidth, gain margin, and phase margin. This modifi-
cation is not convenient for the user and designer.
Figure 3.2 shows the fixed ramp sampling the output voltage in current-mode
control circuit for buck converter. S1 and S2 are the switches, VUG is upper control
signal to drive the switch S1, VLG is lower control signal to drive the switch S2, L is
the output inductor, and RCO is the ESR of the output capacitor CO. The diagram

© Springer Nature Singapore Pte Ltd. 2018 81


W.-W. Chen and J.-F. Chen, Control Techniques for Power
Converters with Integrated Circuit, Power Systems,
https://doi.org/10.1007/978-981-10-7004-4_3
82 3 Designing a Dynamic Ramp with Invariant Inductor …

V
Stable
VCOMP
SE VRAMP

SN
ΔVCS(0) VCS

t
0 DTS TS 2TS

Fig. 3.1 Ramp compensation to prevent subharmonic

S1 VLX L VOUT
VUG IL
VIN
VLG RCO
S2 IOUT
Ri CO
Driver RD1
VCS
VRAMP Ramp
Generator
VSUM
VCOMP VFB
GM

RC VREF RD2

CC

Fig. 3.2 Fixed ramp in current-mode control circuit for buck converter

shows that the output voltage ripple, which includes inductor current information,
can be directly used as the ramp for modulation. The current source IOUT is the
output load current, and RD1 and RD2 are the feedback resistors to determine the
output voltage. The feedback voltage VFB and the reference voltage VREF are built
inside the IC. The ramp generator is always added into the control loop to prevent
subharmonic. RC and CC are the compensator parameters to adjust the systemwloop
stability.
The input voltage rang for low voltage buck is during 2.5–6 V. The input
voltage range for low voltage buck is are optimized for running off single cell
Li-Ion batteries, but can also be used for supplies running from 5 V rails as well.
The low voltage buck can usually regulate the different output voltage rails like 1.2,
1.8, 2.5, and 3.3 V. The low voltage buck is difficult to regulate 5 V output voltage,
3.1 Challenges for Wide Input Voltage Range 83

Table 3.1 Operation conditions in current-mode control circuit for low voltage buck converter
VIN (V) 6V FS (kHz) 400 kHz L (lH) 15 lH
VOUT (V) 3.3 V RD1 (kX) 68 kX CO (lF) 22 lF
IOUT (A) 2A RD2 (kX) 22 kX RCO (mX) 3 mX
VREF (V) 0.8 V Ri (A/V) 0.4

because it is needed to depend on the RDS(ON) of the switch S1 and the maximum
output current rating.
Table 3.1 lists the operation conditions of current-mode control circuit for low
voltage buck converter, where mC is expressed by Eqs. (3.1) and (3.2). SE is a slope
of the ramp compensation and SN is the inductor on-time slope and D′ = 1 − D.
Based on Table 3.1, the open-loop control-to-output bode plot with different mC
values is plotted in Fig. 3.3.

SE
mC ¼ 1 þ ð3:1Þ
SN

VIN  VOUT
SN ¼  Ri ð3:2Þ
L

In Fig. 3.3 [1–5], the gain curves of mC values 1 and 1.2 present no ramp
compensation and small ramp compensation, respectively. Both of these values
have two poles which can cause a high Q at half of the switching frequency. The
two poles cause a sharp phase drop of 180° at a half of the switching frequency,
thus resulting in subharmonic and an unstable system. However, if the ramp
compensation is large, as presented by the gain curve of mC is equal to 8, then the
control loop has formed as a voltage-mode control, given that the ramp compen-
sation is much larger than the inductor current. Therefore, a proper mC value
between 2 and 4 should be designed to prevent subharmonic and to ensure the
stability of the system.
Most low voltage buck converters provide a recommended component selection
table to maintain parameter mC in a proper range, as listed in Table 3.2 [17, 18]. In
order to meet the fixed ramp compensation requires that the inductance is changed
to maintain the same current ripple when the user determines to apply different the
output voltage. Thus, the compensation circuit should also be modified to keep the
system stability. Based on the recommended component selection table, the
inductance is a minor change at different output voltage, because the voltage drop
between the input voltage and output voltage cannot result in a large value for SN.
Table 3.3 lists the same inductance and compensator under different output
voltages in current-mode control circuit for low voltage buck converter. Based on
the conditions listed in Table 3.3, bode plot is constructed to illustrate system sta-
bility, as shown in Fig. 3.4. The operation of the input voltage is 3.3 V for the worst
case scenario (high duty cycle). Inductance and capacitance are also selected at
1.5 lH and 22 lF, respectively, under different output voltages. Figure 3.4 shows
84 3 Designing a Dynamic Ramp with Invariant Inductor …

Fig. 3.3 Open-loop Gain


control-to-output bode plot 20
with different mC
0

-20

dB
mc=1
-40 mc=1.2
mc=2
-60 mc=4
mc=8
-80
100 1000 10000 100000 1000000
Frequency (Hz)
(a) Gain
Phase
0

-50

-100
Degrees

-150 mc=1
mc=1.2
-200
mc=2

-250
mc=4
mc=8
-300
100 1000 10000 100000 1000000
Frequency (Hz)
(b) Phase

Table 3.2 Recommended Conditions VIN = 2.5–6 V, FS = 1000 kHz, IOUT = 2A


component selection for low
IC internal VRAMP = 566 mV, VREF = 827 mV
voltage buck converter
VOUT (V) RD1 (kX) RD2 (kX) L (lH) CO (lF)
3.3 4.3 10 2.2 22
2.5 11.8 10 2.2 22
1.8 20.4 10 1.5 22
1.2 30 10 1.5 22

that the gain curve of the output voltage of 3.3 V still contains current-mode
behavior to increase the loop stability. Based on the control-to-output bode plot
results, the control-to-output cannot contain two poles under different output volt-
ages, so the low voltage buck converter is not a big challenge for avoiding a sub-
harmonic issue, because SN is not a wide range variation.
3.1 Challenges for Wide Input Voltage Range 85

Table 3.3 Same inductance Conditions VIN = 2.5–6 V, FS = 1000 kHz, IOUT = 2A
and compensator under
IC internal VRAMP = 566 mV, VREF = 827 mV
different output voltages for
low voltage buck converter VOUT (V) RD1 (kX) RD2 (kX) L (lH) CO (lF)
3.3 4.3 10 1.5 22
2.5 11.8 10 1.5 22
1.8 20.4 10 1.5 22
1.2 30 10 1.5 22

Fig. 3.4 Open-loop Gain


control-to-output bode plot 10
with different output voltages
at the input voltage 6 V for
low voltage buck converter -5
dB

-20

Vout=3.3V
-35 Vout=2.5V
Vout=1.8V
Vout=1.2V
-50
100 1000 10000 100000 1000000
Frequency (Hz)
(a) Gain

Phase
15

-20

-55
Degrees

-90

-125
Vout=3.3V
-160 Vout=2.5V

-195 Vout=1.8V
Vout=1.2V
-230
100 1000 10000 100000 1000000
Frequency (Hz)
(b) Phase

The wide input voltage range for high voltage buck is during 4.75–22 V. The
wide input voltage range for high voltage buck is normally used for adapter
applications that run from 12 V supply rails or notebook computer applications that
run from 19 V supply rails, but the wide input voltage range makes it possible to
86 3 Designing a Dynamic Ramp with Invariant Inductor …

Table 3.4 Recommended Conditions VIN = 4.75–22 V, FS = 400 kHz, IOUT = 2A


component selection for high
IC internal VRAMP = 306 mV, VREF = 1222 mV
voltage buck converter
VOUT (V) RD1 (kX) RD2 (kX) L (lH) CO (lF)
12 88.7 10 33 22
5 30 10 22 22
3.3 17 10 15 22
1.8 4.75 10 10 22

Table 3.5 Same inductance Conditions VIN = 4.75–22 V, FS = 400 kHz, IOUT = 2A
and capacitance under
IC internal VRAMP = 306 mV, VREF = 1222 mV
different output voltages for
high voltage buck converter VOUT (V) RD1 (kX) RD2 (kX) L (lH) CO (lF)
12 88.7 10 15 22
5 30 10 15 22
3.3 17 10 15 22
1.8 4.75 10 15 22

run them from 5 V rails as well. The high voltage buck can usually regulate the
different output voltage rails like 12, 5, 3.3, and 1.8 V.
Most high voltage buck converters provide a recommended component selection
table to maintain parameter mC in a proper range, as listed in Table 3.4 [17, 18]. In
order to meet the fixed ramp compensation requires that the inductance is changed
to maintain the same current ripple when the user determines to apply different the
output voltage. Thus, the compensation circuit should also be modified to keep the
system stability. This modification is not convenient for the user and designer.
Therefore, this proposed circuit of a novel control scheme can solve this issue.
Table 3.5 lists the same inductance and capacitance under different output
voltages in current-mode control circuit for high voltage buck converter. Based on
the conditions listed in Table 3.5, bode plot is constructed to illustrate system
stability, as shown in Fig. 3.5. The operation of the input voltage is 14 V for the
worst case scenario (high duty cycle). Inductance and capacitance are also selected
at 15 lH and 22 lF, respectively, under different output voltages. Figure 3.5 shows
that the gain curve of the output voltage of 12 V contains two poles which can
result in a high Q at half of the switching frequency. The two poles cause a sharp
phase drop of 180° at a half of the switching frequency, which means that the
system suffers from a subharmonic issue at the output voltage of 12 V, so the high
voltage buck converter is a big challenge for avoiding a subharmonic issue, because
SN has a wide range variation. On the other hand, the compensation circuit is
difficult to design cancelling a subharmonic issue.
3.2 Dynamic Slope Compensation Design 87

Fig. 3.5 Open-loop Gain


control-to-output bode plot 20
with different output voltages
at the input voltage 14 V for 5
high voltage buck converter
-10

dB
-25
Vout=12V
-40
Vout=5V

-55 Vout=3.3V
Vout=1.8V
-70
100 1000 10000 100000 1000000
Frequency (Hz)
(a) Gain
Phase
0

-50

-100
Degrees

-150

-200 Vout=12V

-250 Vout=5V
Vout=3.3V
-300
Vout=1.8V
-350
100 1000 10000 100000 1000000
Frequency (Hz)
(b) Phase

3.2 Dynamic Slope Compensation Design

Figure 3.6 shows the dynamic ramp sampling the output voltage in current-mode
control circuit for buck converter. The ramp generator can only sample the output
voltage or the output voltage and input voltage, to implement a dynamic ramp. The
output voltage is changed large, and the dynamic ramp is increased to prevent the
system suffering from a subharmonic issue.
Figure 3.7 shows that the dynamic ramp is dependent on the output voltage. The
duty cycle D1 is small than 50% as shown in Fig. 3.7a and the duty cycle D2 larger
than 50% as shown in Fig. 3.7b. Based on the duty cycle D1 is smaller than the duty
cycle D2 and the slope of SN1 is steeper than that of SN2. The ramp generator yields
the new ramp compensation VRAMP2 instead of the original ramp compensation
88 3 Designing a Dynamic Ramp with Invariant Inductor …

S1 VLX L VOUT
VUG IL
VIN
VLG RCO
S2 IOUT
Ri CO
Driver RD1
VCS
VRAMP Ramp
VOUT
Generator
VSUM
VCOMP VFB
GM

RC VREF RD2

CC

Fig. 3.6 Dynamic ramp in current-mode control circuit for buck converter

Fig. 3.7 Dynamic ramp V


depended on the output
voltage

VCOMP
SE1 VRAMP1

SN1

VCS
t
0 D1TS TS D1TS 2TS
(a) Duty cycle small than 50%

VCOMP
SE2
VRAMP2

SN2

VCS
t
0 D2TS TS D2Ts 2TS
(b) Duty cycle larger than 50%
3.2 Dynamic Slope Compensation Design 89

Fig. 3.8 Implementation of a VOUT


dynamic ramp generator
sampling the output voltage VCC
R1
G1
VA

R2 VRAMP
VCCS

CLK S C

VRAMP1, which has the slope of SE2 steeper than that of SE1. Thus, ramp com-
pensation is not fixed but dynamic and the dynamic ramp is convenient for the user
and designer, because it does not change the inductor to meet different output
voltages operation.
Figure 3.8 shows the implementation of the dynamic ramp generator sampling
the output voltage. The voltage VA depends on the output voltage. The voltage VA is
also an input signal for the VCCS. The output signal of VCCS charges capacitor
C to generate the dynamic ramp VRAMP, which can be calculated by Eq. (3.3). G1 is
the gain of the voltage-controlled current source. The switch S is controlled to
connect the voltage VRAMP to the ground and the control signal of CLK is usually
used 1–5% of duty cycle to control the switch S.

G1  ðVOUT  R1 Rþ2 R2 Þ
VRAMP ¼ ð3:3Þ
C  TS

Figure 3.9 shows the block diagram of the dynamic ramp sampling the output
voltage in current-mode control circuit for buck converter. In the block diagram, the
switches of the power stage are replaced with a three-terminal switching model [1–
4, 10–13, 19–28], in which the ideal transformer plays the role of the average duty
cycle, and the two dependent sources model the perturbation of the duty cycle.
According to the block diagram, transfer functions can be derived to analyze system
stability. The transfer functions of control-to-output are shown in Eqs. (3.4)–(3.9).
mC should be designed at an optimum proportion between SN and SE to obtain the
suitable signal for SE.

V^OUT ROUT 1
ffi  h i  FpðsÞ  FhðsÞ ð3:4Þ
^ Ri 1 þ 0
LFS ð1 þ SN Þ  D  0:5
VCOMP ROUT SE
90 3 Designing a Dynamic Ramp with Invariant Inductor …

VIN

D L
a c

iˆL iˆOUT
ICd̂ 1 D RCO
ROUT V̂OUT
CO
p

Fm VOUT

Kr V̂OUT

iˆL
HE(s) Ri

FC V̂COMP

Fig. 3.9 Block diagram of the dynamic ramp sampling the output voltage in current-mode control
circuit for buck converter

1 þ s  CO  RCO
FpðsÞ ¼ ð3:5Þ
1 þ xsP

1
FhðsÞ ¼ s þ s2 ð3:6Þ
1þ xn QP x2 n

  
1 1 SE
xP þ  1þ  D0  0:5 ð3:7Þ
CO  ROUT CO  L  FS SN

1
QP ¼ h  i ð3:8Þ
p  1þ SE
SN  D0  0:5

xn ¼ p  FS ð3:9Þ

Based on the conditions listed in Table 3.5 for optimized mC values, inductance
and capacitance are selected at 15 lH and 22 lF, respectively, at the input voltage
of 14 V and the output voltage of 12 V, under different mC values of 3, 6, and 9, as
shown in Fig. 3.10.
3.2 Dynamic Slope Compensation Design 91

Fig. 3.10 Open-loop Gain


control-to-output bode plot 20
with different mC at the input
voltage 14 V and the output
0
voltage 12 V

-20

dB
-40

mc=3
-60 mc=6
mc=9
-80
100 1000 10000 100000 1000000
Frequency (Hz)
(a) Gain
Phase
0

-50

-100
Degrees

-150

-200

-250 mc=3
mc=6
-300
mc=9
-350
100 1000 10000 100000 1000000
Frequency (Hz)
(b) Phase

The gain curve of mC = 3 contains a high Q at half of the switching frequency


caused by two poles. The gain curve of mC = 9 has a voltage mode behavior similar
to having no current loop function. Finally, the gain curve of mC = 6 is optimum for
designing the proportion between SN and SE. This system is stable to operate at a
high duty cycle and also uses the invariant inductor to prevent subharmonic. The
dynamic ramp is designed based on the gain curve of mC = 6. Ramp generator
parameters G1, R1, R2, and C are calculated by Eq. (3.10). Assuming that R1 is
100 kX, R2 is 10 kX, and C is 10 pF, then gain G1 of VCCS is 2.6667l.

G1  ðVOUT  R1 Rþ2 R2 Þ ðVIN  VOUT Þ


5   Ri ð3:10Þ
C L
92 3 Designing a Dynamic Ramp with Invariant Inductor …

Fig. 3.11 Open-loop Gain


control-to-output bode plot of 20
the dynamic ramp sampling
the output voltage with
0
different output voltages at the
input voltage 14 V
-20

dB
-40
Vout=12V
Vout=5V
-60
Vout=3.3V
Vout=1.8V
-80
100 1000 10000 100000 1000000
Frequency (Hz)
(a) Gain
Phase
0

-50
Degrees

-100

-150 Vout=12V
Vout=5V
-200 Vout=3.3V
Vout=1.8V
-250
100 1000 10000 100000 1000000
Frequency (Hz)
(b) Phase

Figure 3.11 shows the open-loop control-to-output bode plot of the dynamic
ramp sampling the output voltage with different output voltages at the input voltage
14 V. The open-loop control-to-output for buck converter using the dynamic ramp
exhibits the same bode plot for the gain and phase under different output voltages.
This system does not only use the invariant inductor, but also designs the same
compensation circuit to meet different applications when the output voltage is
changed. Therefore, the proposed dynamic ramp sampling the output voltage is
very simple and easy to implement.
The ramp generator can also sample the output voltage and the input voltage to
implement the dynamic ramp. The voltage droop between the input voltage and the
output voltage is changed from large to small, and the dynamic ramp is increased to
prevent the system suffering from a subharmonic issue. Figure 3.12 shows the
3.2 Dynamic Slope Compensation Design 93

VIN VOUT

VCC
R1 R3
G1 G2
VA
VB
VRAMP
VCCS
VCVS VREFG
R2 R4
CLK S C

Fig. 3.12 Implementation of a dynamic ramp generator sampling the output voltage and input
voltage

implementation of the dynamic ramp generator with sampling the output voltage
and input voltage.
The voltage VA depends on the input voltage. The voltage VA is also an input
signal for a VCVS. The voltage VB depends on the output voltage. This voltage is
the other input signal for a VCVS. G1 is the gain of VCVS. G1 is also a unity gain.
The output signal of VCVS is an input signal for VCCS. G2 is the gain of VCCS.
The output signal of VCCS charges capacitor C to generate the dynamic ramp
VRAMP, which can be calculated by Eq. (3.11). Otherwise, the voltage signal VREFG
should be designed to be larger than the voltage droop between the VA and VB.
h i
G2  VREFG  ðVIN  R1 Rþ2 R2  VOUT  R3 Rþ4 R4 Þ
VRAMP ¼ ð3:11Þ
C  TS

Figure 3.13 shows the block diagram of the dynamic ramp with sampling the
output voltage and the input voltage in current-mode control circuit for buck con-
verter. Finally, the gain curve of mC = 6 is also optimum for designing the pro-
portion between SN and SE as shown in Fig. 3.10. This system is stable to operate at a
high duty cycle and also uses the invariant inductor to prevent subharmonic. Ramp
generator parameters G2, R1, R2, R3, R4, VREFG, and C are calculated by Eq. (3.11).
Figure 3.14 shows the open-loop control-to-output bode plot of the dynamic
ramp sampling the output voltage and the input voltage with different output
voltages at the input voltage of 14 V. According to the same bode plot, this system
does not only use the invariant inductor, but also designs the same compensation
circuit to meet different applications when the output voltage is changed. Therefore,
the proposed dynamic ramp sampling the output voltage and the input voltage is
very simple and easy to implement.
94 3 Designing a Dynamic Ramp with Invariant Inductor …

VIN

D L
a c

iˆL iˆOUT
ICd̂ RCO

D
1
ROUT V̂OUT
CO
p


VIN Fm VOUT

kr V̂OUT

iˆL
He( s ) Ri

Fc VˆCOMP

Fig. 3.13 Block diagram of the dynamic ramp sampling the output voltage and the input voltage
in current-mode control circuit for buck converter

3.3 Experimental Results

MathCAD predictions, SIMPLIS simulation results, and experimental verifications


were conducted to determine the feasibility and performance of the proposed
dynamic ramp with the invariant inductor in current-mode control circuit for buck
converter. The specifications are as follows:
1. Input DC voltage range (VIN): 14 V
2. Output DC voltage range (VOUT): 12 V
3. Output load current (IOUT): 2 A
4. Switching frequency (FS): 400 kHz
5. Feedback resistors (RD1, RD2): 115 and 13 kX
6. Main inductor (L): 15 lH
7. Output capacitors (CO): 22 lF/ 25 V * 1 (RCO: 3 mX)
8. Reference voltage (VREF): 1.222 V
3.3 Experimental Results 95

Fig. 3.14 Open-loop Gain


control-to-output bode plot of 20
the dynamic ramp sampling
the output voltage and the 0
input voltage with different
output voltages at the input
voltage 14 V -20

dB
-40
Vout=12V
Vout=5V
-60
Vout=3.3V
Vout=1.8V
-80
100 1000 10000 100000 1000000
Frequency (Hz)
(a) Gain

Phase
0

-50
Degrees

-100

-150 Vout=12V
Vout=5V
-200 Vout=3.3V
Vout=1.8V
-250
100 1000 10000 100000 1000000
Frequency (Hz)
(b) Phase

9. Dynamic ramp circuit (Fig. 3.8): G1 = 2.6667 lV/A; C = 10p; R1 = 100 kX;
R2 = 10 kX
10. Fixed ramp circuit: VRAMP = 0.306 V
11. Compensator circuit: RC = 10 kX; CC = 1.5 nF
Figure 3.15 compares MathCAD predictions and SIMPLIS simulation results
for the open-loop control-to-output bode plot using the dynamic ramp and the fixed
ramp. MathCAD is used for predicting results in calculating equations for the
dynamic ramp and the fixed ramp. SIMPLIS is used to simulate results for the
dynamic ramp and the fixed ramp. MathCAD prediction results are very similar to
SIMPLIS simulation results. The fixed ramp has a high Q at half of the switching
frequency caused by two poles. The two poles cause a sharp phase drop of 180° at
200 kHz, which is also located at half of the switching frequency, thus resulting in a
subharmonic issue and an unstable system.
96 3 Designing a Dynamic Ramp with Invariant Inductor …

Fig. 3.15 Comparison of Gain


MathCAD predictions and 20
SIMPLIS simulation results
for open-loop 0
control-to-output bode plot
using the dynamic ramp and
the fixed ramp -20

dB
-40
Dynamic Ramp Predictions
-60 Dynamic Ramp Simulaiton
Fixed Ramp Predictions
Fixed Ramp Simulaiton
-80
1000 10000 100000 1000000
Frequency (Hz)
(a) Gain
Phase
0

-50

-100
Degrees

-150

-200
Dynamic Ramp Predictions
-250
Dynamic Ramp Simulaiton
-300 Fixed Ramp Predictions
Fixed Ramp Simulaiton
-350
1000 10000 100000 1000000
Frequency (Hz)
(b) Phase

Figure 3.16 compares MathCAD predictions and SIMPLIS simulation results


for the closed-loop loop gain bode plot using the dynamic ramp and the fixed
ramp. In Fig. 3.15, MathCAD prediction results are very similar to SIMPLIS
simulation results. Both MathCAD prediction and SIMPLIS simulation result
verifications confirm that the proposed dynamic ramp with the invariant inductor in
current-mode control circuit for buck converter can significantly prevent subhar-
monic. The bandwidth of the fixed ramp for the SIMPLIS simulation results and
MathCAD predictions are both 20 kHz; whereas the bandwidth of the dynamic
ramp for SIMPLIS simulation results and MathCAD predictions are very close to
each other. The phase margin of the dynamic ramp is very close to the fixed ramp.
Figure 3.17 shows the chip layout of the control IC. The control IC was used by
TSMC 0.6 lm process and RDS(ON) of the main switch is 180 mX. The Power
3.3 Experimental Results 97

Fig. 3.16 Comparison of Gain


MathCAD predictions and 60
SIMPLIS simulation results
for closed-loop loop gain 40
bode plot using the dynamic
20
ramp and the fixed ramp
0

dB
-20
Dynamic Ramp Predictions
-40
Dynamic Ramp Simulaiton
-60 Fixed Ramp Predictions
Fixed Ramp Simulaiton
-80
1000 10000 100000 1000000
Frequency (Hz)
(a) Gain

Phase
0

-50

-100
Degrees

-150

-200
Dynamic Ramp Predictions
-250 Dynamic Ramp Simulaiton
-300 Fixed Ramp Predictions
Fixed Ramp Simulaiton
-350
1000 10000 100000 1000000
Frequency (Hz)
(b) Phase

MosFET, driver circuit, control circuit, and current sensor occupy the marked area
with a die size of 1250 lm  960 lm.
Figure 3.18 shows the output voltage and the VUG signal using the fixed ramp at
the input voltage of 14 V, the output voltage of 12 V, and the output load current of
2A. Experimental results are measured at the output voltage and the VUG signal for
the fixed ramp. The experimental result of the output voltage ripple for the fixed
ramp is equal to 27.2 mV. The VUG signal for the fixed ramp exhibits subharmonic,
thus the width of the VUG signal is not the same.
Figure 3.19 shows the output voltage and the VUG signal using the dynamic
ramp at the input voltage of 14 V, the output voltage of 12 V, and the output load
current of 2 A. Experimental results are measured at the output voltage and the VUG
signal for the dynamic ramp. The experimental result of the output voltage ripple
98 3 Designing a Dynamic Ramp with Invariant Inductor …

Power MosFET

960 μm
OSC Driver
Current Sense

Control Regulator

Bias GM

1250 μm

Fig. 3.17 Chip layout of the control IC

Fig. 3.18 VOUT and VUG


signals using the fixed ramp at
the input voltage 14 V and the VUG(10V/2μs)
output voltage 12 V

(1)

VOUT(10mV/2μs)
27.2mV

(2)

(1),(2) With Fixed Ramp of VOUT and VUG

for the dynamic ramp is equal to 6.8 mV. The VUG signal does not exhibit sub-
harmonic and the width of the VUG signal is the same.
Figure 3.20 compares of the experimental results for efficiency using the
dynamic ramp and the fixed ramp at the input voltage 14 V and the output voltage
3.3 Experimental Results 99

Fig. 3.19 VOUT and VUG


signals using the dynamic
ramp at the input voltage VUG(10V/2μs)
14 V and the output voltage
12 V
(1)

VOUT(10mV/2μs)

6.8mV (2)

(1),(2) With Dynamic Ramp of VOUT and VUG

Fig. 3.20 Comparison of the 100


experimental results for
90
efficiency using the dynamic (1)
ramp and the fixed ramp at the 80
input voltage 14 V and the 70
output voltage 12 V (2)
Efficiency (%)

60

50

40

30

20
(2) With Fixed Ramp
10
(1) With Dynamic Ramp
0
0.001 0.01 0.1 1 10
Output Load (A)

12 V in current-mode control circuit for buck converter. The current-mode control


circuit using the dynamic ramp for buck converter has higher efficiency than the
fixed ramp during the output load current 1 mA to 2 A, because the fixed ramp has
the subharmonic resulting in the large switching loss. When the output load current
has increased, the efficiency of that with the fixed ramp is very close to that with the
dynamic ramp, because the conduction loss is a major factor that impacts the
efficiency in this condition.
Figure 3.21 compares MathCAD predictions and experimental results for the
closed-loop loop gain bode plot using the dynamic ramp and the fixed
ramp. MathCAD prediction results are very similar to experimental results. Both
MathCAD prediction and experimental result verification confirm that the proposed
100 3 Designing a Dynamic Ramp with Invariant Inductor …

Fig. 3.21 Comparison of Gain


MathCAD predictions and 60
experimental results for
40
closed-loop loop gain bode
plot using the dynamic ramp 20
and the fixed ramp
0

dB
-20
Dynamic Ramp Predictions
-40
Dynamic Ramp Experimenta
-60 Fixed Ramp Predictions
Fixed Ramp Experimental
-80
1000 10000 100000 1000000
Frequency (Hz)
(a) Gain

Phase
0

-50

-100
Degrees

-150

-200
Dynamic Ramp Predictions
-250
Dynamic Ramp Experimenta
-300 Fixed Ramp Predictions
Fixed Ramp Experimental
-350
1000 10000 100000 1000000
Frequency (Hz)
(b) Phase

dynamic ramp with the invariant inductor in current-mode control circuit for buck
converter can significantly prevent subharmonic. The gain curve of the fixed ramp
contains a high Q at 200 kHz, which is also located at half of the switching
frequency caused by two poles.
The dynamic ramp and fixed ramp both uses the same compensation parameters
to obtain MathCAD predictions and experimental results. The bandwidth of the
fixed ramp for the experimental results and MathCAD predictions are both 20 kHz;
whereas the bandwidth of the dynamic ramp for the experimental results and
MathCAD predictions are very close to each other. The phase margin of the
dynamic ramp is very close to the fixed ramp for both the experimental results and
MathCAD predictions.
3.4 Summary 101

3.4 Summary

Experimental results, SIMPLIS simulation results, and MathCAD predictions


confirm that the proposed dynamic ramp with the invariant inductor in
current-mode control circuit for buck converter does not only maintain system
stability under different input/output voltages without changing the inductor and
compensation circuit, but also significantly prevents subharmonic. Moreover, the
proposed dynamic ramp has a very simple structure and easy-to-implement design.
This system is available and useful as a current-mode control for buck converter.

References

1. R.B. Ridley, A new, continuous-time model for current-mode control. IEEE Trans. Power
Electron. 6, 271–280 (1991)
2. R.B. Ridley, A new continuous-time model for current-mode control with constant on-time,
constant off-time, and discontinuous conduction mode, in Proceedings on IEEE Power
Electronics Specialists Conference (1990), pp. 382–389
3. J. Li, F.C. Lee, New modeling approach and equivalent circuit representation for
current-mode control. IEEE Trans. Power Electron. 1218–1230 (2010)
4. R.W. Erickson, D. Maksimovic, Fundamentals of Power Electronics (Kluwer, Norwell, MA,
2001)
5. R.D. Middlebrook, S. Cuk, A general unified approach to modeling switching-converter
power states, in Proceedings on IEEE Power Electronics Specialists Conference (1976),
pp. 73–86
6. N. Mohan, W.P. Robbins, P. Imbertson, T.M. Undeland, R.C. Panaitescu, A.K. Jain, P. Jose,
T. Begalke, Restructuring of first courses in power electronics and electric drives that
integrates digital control. IEEE Trans. Power Electron. 18, 429–437 (2003)
7. N. Mohan, Power electronics circuits: An overview, in Proceedings on IEEE Industrial
Electronics Society Conference (1988), pp. 522–527
8. W.H. Ki, Signal flow graph in loop gain analysis of dc-dc PWM CCM switching converters.
IEEE Trans. Circ. Syst.-I 6, 644–654 (1998)
9. W.H. Ki, Analysis of subharmonic oscillation of fixed-frequency current-programming switch
mode power converters. IEEE Trans. Circ. Syst.-I 45(1), 104–108 (1998)
10. J. Li, Current-mode control: modeling and its digital application, PHD thesis, Virginia
Polytechnic Institute and State University, 2009
11. W.W. Chen, J.F. Chen, T.J. Liang, L.C. Wei, W.Y. Ting, Designing a dynamic ramp with an
invariant inductor in current-mode control for an on-chip Buck converter. IEEE Trans. Power
Electron. 29(2), 750–758 (2014)
12. W.W. Chen, J.F. Chen, T. J. Liang, L. C. Wei, W.Y. Ting, Dynamic Ramp with the invariant
inductor in current-mode control for Buck converter, in Proceedings on IEEE Applied Power
Electronics Conference and Exposition (APEC) (2013), pp. 1244–1249
13. W.W. Chen, J.F. Chen, T.J. Liang, Dynamic Ramp control in current-mode adaptive on-time
control for Buck converter on chip, in Proceedings on IEEE Future Energy Electronics
Conference and ECCE Asia (IFEEC 2017—ECCE Asia) (2017), pp. 280–285
14. F. Tian, S. Kasemsan, I. Batarseh, An adaptive slope compensation for the single-stage
inverter with peak current-mode control. IEEE Trans. Power Electron. 2857–2862 (2011)
15. Z. Zansky, Current Mode Converter with Controlled Slope Compensation. United States
Patent, Patent Number: 4,837,495, Date of Patent: June 6, 1989
102 3 Designing a Dynamic Ramp with Invariant Inductor …

16. L. Yanming, L. Xinquan, C. Fuji, Y. Bing, J. Xinzhang, An adaptive slope compensation


circuit for buck DC-DC converter, in Proceeding of the International Conference on ASIC
2007, ASICON’07 (Oct 2007) pp. 608–611
17. Richtek Tech. Corp., 2A, 22 V, 400 kHz Step-Down Converter. RT8267 Datasheet (2011)
18. Richtek Tech. Corp., 2A, 23 V, 340 kHz Synchronous Step-Down Converter. RT8294
Datasheet (2011)
19. R. Redl, N.O. Sokal, Current-mode control, five different types, used with the three basic
classes of power converters: Small-signal ac and large-signal dc characterization, stability
requirements, and implementation of practical circuits, in Proceedings on IEEE Power
Electronics Specialists Conference (1985), pp. 771–785
20. N. Lakshminarasamma, M. Masihuzzaman, V. Ramanarayanan, Steady-state stability of
current-mode active-clamp ZVS DC-DC converters. IEEE Trans. Power Electron. 26(5),
1295–1304 (2011)
21. B. Bryant, M.K. Kazimierczuk, Modeling the closed-current loop of PWM dc-dc converters
operating in CCM with peak current-mode control. IEEE Trans. Circ. Syst.-I 52(11), 2404–
2412 (2005)
22. N. Kondrath, M.K. Kazimierczuk, Control current and relative stability of peak current-mode
controlled pulse-width modulated dc-dc converters without slope compensation. IET Power
Electron. 3(6), 936–946 (2010)
23. M.K. Kazimierczuk, Pulse-Width Modulated DC-DC Power Converters (Wiley, 2008)
24. V. Vorperian, Simplified analysis of PWM converters using model of PWM switch-I.
Continuous conduction mode. IEEE Trans. Aerosp. Electron. Syst. 26(3), 490–496 (1990)
25. V. Vorperian, Simplified analysis of PWM converters using model of PWM switch-II.
Discontinuous conduction mode. IEEE Trans. Aerosp. Electron. Syst. 26(3), 490–496 (1990)
26. B. Bryant, M.K. Kazimierczuk, Open-loop power-stage transfer functions relevant to
current-mode control of boost PWM converter operating in CCM. IEEE Trans. Circ. Syst.-I
52(11), 2158–2164 (2005)
27. M.M. Jovanovic, L. Huber, Small-signal modeling of nonideal magamp PWM switch. IEEE
Trans. Power Electron. 14(5), 882–889 (1999)
28. Y. Yingyi, F.C. Lee, P. Mattavelli, Unified three-terminal switch model for current mode
controls. IEEE Trans. Power Electron. 27(9), 4060–4070 (2012)
Chapter 4
Review of the Adaptive On-time Control
Circuits for Buck Converters

4.1 Adaptive On-time Control Circuits for Buck


Converters

Many novel control circuits, such as CPU and electronic devices, have been
reported for power supplies to meet stringent requirements in recent years. These
devices can reduce standby power loss and increase the load transient response to
achieve high performance and low loss of system design. Owing to the rapid
development of microprocessors, over a billion transistors have been integrated into
one processor. Core static current has been increased from 20 to 100 A, and core
voltage has been reduced from 2 to 0.7 V [1–3]. Moreover, CPU load transient may
occur within 1 µs. These requirements pose a stringent challenge to voltage reg-
ulation (VR) [4–8], especially in the load transient response. The use of many
output capacitors to reduce the voltage spike during transient is an approach that has
been adopted. For future microprocessors, increasing the number of capacitors to
meet even higher transient requirements will not be acceptable because of size and
cost issues.
The power converter must be able to regulate its output voltage to be near
constant as the load current demand varies anywhere from zero to full load, even
when the change occurs in a relatively short time. A good performance of load
transient response can save on output capacitor size and cost. Meanwhile, settling
time and stability can be displayed in the load transient response, so power con-
verter performance must be tested. Based on these requirements, the conventional
COT control circuit is widely used in CPU applications and other electronic devices
with high slew rates because of the advantages of faster load transient response and
better light-load efficiency compared with the current-mode control. The first reason
is that the conventional COT control circuit does not have a compensator, which
delays the system loop response. The second reason is that the conventional COT
control circuit is a type of PFM, so it can generate a minimum off-time at load
droop, which is very useful in preventing the output voltage from dropping

© Springer Nature Singapore Pte Ltd. 2018 103


W.-W. Chen and J.-F. Chen, Control Techniques for Power
Converters with Integrated Circuit, Power Systems,
https://doi.org/10.1007/978-981-10-7004-4_4
104 4 Review of the Adaptive On-time Control …

significantly. Thus, the conventional COT control circuit is suitable for use in fast
load transient power supply applications.
However, the load transient response of conventional COT control circuit
requires further improvement to save output capacitance, so it is used in CPU
application, but CPU load transient may occur within 1 µs, so the conventional
COT control circuit is also needed to add a non-linear open loop quick response
circuit to improve transient response [9–21]. Further, most conventional COT
controls provide voltage to CPU, try to solve this issue by setting up drop-voltage
thresholds to trigger another open loop regulation mechanism, such as triggering
another on-time generator or increasing the duty of its on-time generator. However,
this kind of design has two major drawbacks. First, the threshold is discrete, which
means it may improve transient response over a specific threshold. Second, the
threshold is fixed, which cannot meet the variety of loading conditions. Moreover,
if the threshold can be set by external components, it suffers another drawback of
adding extra pins, which increases cost and reduces the flexibility of board design.
Our desire is to provide a method, which can maintain a switching regulator’s
loop stability and characteristic, and also provide a nearly real-time boost response
to help the regulator trace back the correct voltage. This method can allow the
output voltage drop to decide dynamically how fast or what quantity to boost the
regulator without the extra setting of pins and provide board designers the flexibility
to change boost speed.
Circuit diagram of the conventional COT control circuit for buck converter is as
shown Fig. 4.1. SA and SB are the switches, L is the output inductor, and RCO is the
ESR of the output capacitor CO. The current source IOUT is the output load current,

VIN SA L VOUT
VLX
VUG
IL IOUT
VLG RCO
SB
CO
RD1
Driver

VFB
On-Time VTRIG
Generator
VREF RD2

Fig. 4.1 Circuit diagram of the conventional COT control circuit for buck converter
4.1 Adaptive On-time Control Circuits for Buck Converters 105

VIN SA VLX L VOUT


VUG
IL IOUT
VLG RCO
SB
CO
RD1
Driver

VFB
On-Time VTRIG
Generator
VREF RD2

VIN VOUT

Fig. 4.2 Adaptive on-time control circuits for buck converter

and RD1 and RD2 are the feedback resistors to determine the output voltage. VFB is a
feedback voltage. The reference voltage VREF is created inside the IC.
The on-time generator circuit can generate the fixed on-time width to control the
driver circuit and achieve the voltage regulation if the conventional COT control
circuit wants to regulate a high VOUT and an increase in switching loss occurs.
The adaptive on-time control circuits sample the VIN signal and VOUT signal to
adjust the on-time width to control the driver circuit and achieve the voltage reg-
ulation. The adaptive on-time control circuits for buck converter are as shown
Fig. 4.2. SA and SB are the switches, L is the output inductor, and RCO is the ESR of
the output capacitor CO. The current source IOUT is the output load current, and RD1
and RD2 are the feedback resistors to determine the output voltage. VFB is a feed-
back voltage. The reference voltage VREF is created inside the IC. The adaptive
on-time control circuits sample the VIN and VOUT signals to adjust the on-time width
to control the driver circuit and achieve voltage regulation. The adaptive on-time
control circuits for buck converter are shown Fig. 4.2. The on-time generator circuit
samples the VIN and VOUT signals to adjust the on-time width to control the driver
circuit and achieve voltage regulation. Two types of adaptive on-time control cir-
cuits based on connection form are typically used, namely, constant frequency
on-time control circuit and constant current ripple on-time control circuit [18–22].
In adaptive on-time control circuits, the absence of virtual inductor current ripple
to add to feedback voltage VFB causes time-delay effects in the loop. Thus, these
control circuits suffer from subharmonic oscillations, especially when low-ESR
capacitors, such as ceramic capacitors, are used [18–32]. The transfer function of
control-to-output GCTO(s) is simplified by Eqs. (4.1)–(4.5) [31, 32].
106 4 Review of the Adaptive On-time Control …

1 ðRCO  CO  s þ 1Þ
GCTO ðs)   ð4:1Þ
1 þ s=ðx1  Q1 Þ þ s2 =x21 1 þ s=ðx2  Q3 Þ þ s2 =x22
p
x1 ¼ ð4:2Þ
TON

2
Q1 ¼ ð4:3Þ
p
x2 ¼ p  Fs ð4:4Þ

1
Q3 ¼   ð4:5Þ
p  RCO  CO  TON
2  Fs

The transfer function indicates that the double pole at half of the switching
frequency may move to the right half-plane according to the different capacitor
parameters. The critical condition for stability is obtained by Eq. (4.6), which
clearly shows the influence of capacitance ripple [18–32].

TON
RCO  CO [ ð4:6Þ
2

For real capacitors, under the condition 267 kHz of switching frequency, 12 V
of input voltage, and 3.3 V of output voltage, the parameter of a single POSCAP
capacitor is 330 lF, and its ESR of 4.5 mX meets Eq. (4.6). Thus, the system is
stable. However, the parameters of the five ceramic capacitors with 22 lF and ESR
of 3 mX in parallel do not meet Eq. (4.6), so sub-harmonic oscillation occurs.
Experimental results for output voltage and VUG signal with subharmonic
oscillation using the ceramic capacitors and without subharmonic oscillation using
POSCAP capacitor are as shown in Fig. 4.3. The experimental waveforms include
the VUG and VOUT signals, both with and without subharmonic oscillation. The
subharmonic oscillation can directly cause very high output voltage ripple with
double pulses or multiple pulses. The output voltage with subharmonic oscillation
is 284 mV with double pulses higher than that without subharmonic oscillation
(67 mV), indicating a difference of about four times ripple magnitude difference in
the experimental waveforms. Besides, the switching loss with subharmonic oscil-
lation is larger than that without subharmonic oscillation [18–32].
The subharmonic oscillation can result a large output voltage and the large
output voltage may damage the output loading equipment or devices or make the
output loading equipment or devices to suffer an error operation. On the other hand,
the subharmonic oscillation is also an impact on the load efficiency.
Figure 4.4 shows a comparison of the efficiency of the experimental results with
subharmonic oscillation using ceramic capacitors and without subharmonic oscil-
lation using POSCAP capacitor [18–22, 31–35]. The subharmonic oscillation can
4.1 Adaptive On-time Control Circuits for Buck Converters 107

VUG (20V/4μs)

VOUT (200mV/4μs) 67mV

VUG (20V/4μs)

VOUT (200mV/4μs) 284mV

(2),(4) With Subharmonic Oscillation


(1),(3) Without Subharmonic Oscillation

Fig. 4.3 Comparison of experimental results for output voltage and VUG signal with subharmonic
oscillation using ceramic capacitors and without subharmonic oscillation using POSCAP capacitor

Fig. 4.4 Comparison of the 100


efficiency of the experimental (1)
90
results with subharmonic
oscillation using ceramic 80
(2)
capacitors and without 70
Efficiency (%)

subharmonic oscillation using 60


POSCAP capacitor
50
40
30
20
10 (2)With Subharmonic
(1) Without SubharmonicOscillation
Oscillation
(1)Without Subharmonic
(2) With Subharmonic Oscillation
Oscillation
0
0.001 0.01 0.1 1 10 100
Output Current (A)

directly cause large switching loss. The difference between the efficiency of the
converter that has subharmonic oscillation (43.45%) and the one without subhar-
monic oscillation (54.4%) is about 11% at a light load of 0.2 A. However, the
efficiency of that with subharmonic oscillation is very close to that without sub-
harmonic oscillation at a heavy load of 20 A, because the conduction loss is a major
factor that affects the efficiency in this condition.
108 4 Review of the Adaptive On-time Control …

4.2 Ripple-Based Adaptive On-time Control Circuits


with Virtual Inductor Current Ripple for Buck
Converters

However, in many applications such as smart phone, netbook, and tablet of con-
sumer products, ceramic capacitors are preferred due to its small size, low output
voltage ripple, low cost, and high reliability requirements. One main solution to
eliminate subharmonic oscillation is discussed in [18–22, 31–35]. One control
circuit of ripple-based adaptive on-time control circuit is by adding a victual
inductor current ramp or the internal ramp generator.
Ripple-based adaptive on-time control circuit with virtual inductor current ripple
for buck converter is as shown in Fig. 4.5. SA and SB are the switches, L is the
output inductor, and RCO is the ESR of the output capacitor CO. The current source
IOUT is the output load current, and RD1 and RD2 are the feedback resistors to
determine the output voltage. VFB is a feedback voltage. The reference voltage VREF
is created inside the IC. The on-time generator circuit samples the VIN signal and
VOUT signal to adjust the on-time width to control the driver circuit and achieve the
voltage regulation.

SA VLX L
VIN VOUT
VUG
IL IOUT
VLG SB Rco
Co
Driver
RD1

VSUM VFB
VTRIG
- K1
On-Time
Generator +
VREF RD2

VIN VOUT K2

VIC

VLX RLPF VCLPF VIC

CLPF
DC value
extractor

Fig. 4.5 Ripple-based adaptive on-time control circuit with virtual inductor current ripple for
buck converter
4.2 Ripple-Based Adaptive On-time Control Circuits with … 109

The use of a virtual inductor current ripple is proposed to alleviate the instability
problem because it enhances the effect of the ESR voltage ripple in the feedback
voltage. This circuit provides better system stability, especially in all ceramic output
capacitors case, which normally has relatively low ESR values. In a virtual inductor
current ripple generator, the inductor current ripple waveform can be obtained by
integrating the VLX terminal voltage through RLPF and CLPF integrator and
removing its DC value by DC value extractor. The DC value of VCLPF is removed
by the DC value extractor to generate a virtual inductor current ripple. The virtual
inductor current ripple is added in the feedback voltage VFB to enhance the ESR
voltage ripple [18–22, 31–35].
The operation waveforms of the virtual inductor current can be explained by the
detailed waveforms shown in Fig. 4.6. The rising slope MRISE and falling slope
MFALL of VCLPF can be derived by Eqs. (4.7)–(4.8), respectively. The corner fre-
quency of the low-pass filter of RLPF and CLPF just should be designed close to the
resonant frequency of the output inductor L and output capacitor CO, such that the
current flowing through RLPF is a nearly constant value to charge or discharge
capacitor CLPF during on-time and off-time.

VIN  VOUT
MRISE ¼ ð4:7Þ
RLPF  CLPF
V OUT
M FALL ¼ ð4:8Þ
RLPF  CLPF

It can be seen from Eqs. (4.7)–(4.8) and Fig. 4.7 that virtual ripple VCLPF is
proportional to steady-state ESR voltage ripple. The ESR ripple component of the

Fig. 4.6 Operation V


waveforms of the virtual
inductor current
On-Time

(0)
t

VLX (VIN)

(0)
t
VCLPF
(VOUT)
t
V -V - VOUT
M RISE = IN OUT M FALL =
RLPF × C LPF RLPF × CLPF
110 4 Review of the Adaptive On-time Control …

VRCO (0)
t

VCO [(1+RD1/RD2)*VREF]

t
VOUT
[(1+RD1/RD2)*VREF]

t
VCLPF
(VOUT)

t
VSUM
(VFB)

t
VLX (VIN)

(0)
t

Fig. 4.7 Waveforms of the virtual inductor current ripple generator

negative modulator input VSUM is effectively enhanced by the addition of virtual


ripple VCLPF that makes the system more stable [18–22, 31–35].
The virtual inductor current ripple provides good system stability without the
sensing current information or adding extra components. If the control circuit can
be embedded by a real IC implemented circuit and the signal of the VLX terminal
exists in IC, then no extra pin of IC is needed to implement the modified control.
Finally, based on similar operation conditions, a comparison is made between
ripple-based adaptive on-time control circuit with virtual inductor current ripple and
adaptive on-time control circuit to understand the advantage and superiority of the
elimination of subharmonic oscillation. The operation conditions are 267 kHz of
switching frequency, 12 V of input voltage, 0.765 V of reference voltage, and
3.3 V of output voltage with no load. The ceramic capacitors are 50 lF, and the
ESR of 1 mX does not meet Eq. (4.6).
Figure 4.8a shows the simulation results at the steady status for the adaptive
on-time control circuit. The red-colored waveform of the output voltage signal is
4.2 Ripple-Based Adaptive On-time Control Circuits with … 111

approximately 1.4 V with subharmonic oscillation, and the blue-colored waveform


of the VRCO signal is approximately 30 mV of the peak-to-peak voltage. Feedback
voltage VFB without the virtual inductor current ripple causes time-delay effects in
the loop. Figure 4.8b shows the simulation results at the steady status for the
ripple-based adaptive on-time control circuit with virtual inductor current ripple.
The red-colored waveform of the output voltage signal is approximately 90 mV
without subharmonic oscillation, and the blue-colored waveform of the VRCO signal
is approximately 9 mV of the peak-to-peak voltage because the ESR is 1 mX. Even
the output voltage is without subharmonic oscillation. The virtual inductor current
ripple is added to feedback voltage VFB, and the ripple-based adaptive on-time
control circuit with virtual inductor current ripple is without time-delay effects in
the loop. Thus, the ripple-based adaptive on-time control circuit with virtual
inductor current ripple can use ceramic capacitors to reduce the bill of material
(BOM) size and achieve a small output voltage ripple.

4.3 Current-Mode Adaptive On-time Control Circuit


for Buck Converters

In the adaptive on-time control topology, the absence of inductor current ripple to
be added to the capacitor ripple [18–22, 31–35], which is the integration of AC
ripples in the inductor current, causes time-delay effects in the loop. Hence, this
control scheme suffers from instability caused by subharmonic oscillations because
converters with low ESR, such as ceramic capacitors, are used. The current-mode
adaptive on-time control circuit for buck converter can be implemented to apply
ceramic capacitors with a low ESR because reference voltage VREF and feedback
voltage VFB serve as the input terminals of an error amplifier for the current-mode
adaptive on-time control circuit and not as the input terminals of a comparator.
Thus, the current-mode adaptive on-time control circuit has better noise immunity
than the adaptive on-time control circuit.
Figure 4.9 shows the circuit diagram of a current-mode adaptive on-time control
circuit for buck converter [36]. In the diagram, SA and SB are the switches, L is the
output inductor, DCR is the DC resistor of the inductor, Ri is a current sense ratio
that is used to transfer an inductor-current signal to a voltage signal VCS, and RCO is
the ESR of output capacitor CO. The diagram shows that the output voltage ripple,
which includes inductor-current information, can be directly used as the ramp for
modulation. Current source IOUT is the output load current, and RD1 and RD2 are the
feedback resistors that determine the output voltage. The driver circuit uses the
input signal on-time width to generate two output signals, VUG and VLG; these two
signals should not be turned on at the same time, because such operation causes the
system to have a shoot-through problem. The compensation of RC and CC should
have an optimal design to increase the transient response, if the system simply
connects RC, and it makes the system operate at load line status. Only the feedback
112 4 Review of the Adaptive On-time Control …

(a)

VCO
(V)

VRCO
(mV)

VFB
(V)
VREF
(V)

VOUT
(V)

t (µs)

(b)

VCO
(V)

VRCO
(mV)

VFB
(V)
VREF
(V)

VOUT
(V)

t (µs)

Fig. 4.8 a Simulation results at the steady status for adaptive on-time control circuit. b Simulation
results at the steady status for ripple-based adaptive on-time control circuit with virtual inductor
current ripple
4.3 Current-Mode Adaptive On-time Control Circuit … 113

VIN SA VLX L DCR VOUT


VUG
IL IOUT
VLG RCO
SB
CO
RD1
Driver
Ri
VCS VFB
On-Time VCMP VCOMP
EA
Generator VRAMP
VREF RD2
RC
Ramp
VIN VOUT Generator CC

Fig. 4.9 Current-mode adaptive on-time control circuit for buck converter

signal VFB and reference voltage VREF are built inside the IC. The output signal
VCMP of the comparator depends on the results on the input signals VCOMP, VCS, and
VRAMP.
The current-mode adaptive on-time control circuit has an error amplifier, and the
output terminal of the error amplifier can amplify a voltage different between VFB
and VREF. Thus, it can be used to avoid subharmonic oscillation with double or
multiple pulses even when the output capacitor uses ceramic capacitors. The circuit
diagram of the current-mode adaptive on-time control circuit is similar to that of the
current-mode control circuit. The current-mode adaptive on-time control circuit also
needs to sense the inductor current to increase its system loop response. It produces
a subharmonic issue when the duty cycle is larger than 50%. A ramp generator is
always added to the control loop to prevent the subharmonic issue [36], as shown in
Fig. 4.10. The ramp generator is designed according to the inductor current ripple,
which is determined by the input voltage, output voltage, and inductance. A general
power control IC often embeds a fixed ramp into the control loop, and this ramp can
directly increase noise immunity and prevent the system from suffering from a
subharmonic issue [36]. Voltage VA depends on constant voltage VCC, which is also
an input signal for VCCS. The output signal of VCCS charges capacitor C to
generate fixed ramp VRAMP, which can be calculated using Eq. (4.9).

G1  ðVCC  R1 þ R2 Þ
R2
VRAMP ¼ ð4:9Þ
C  TS

Under similar operation conditions, a comparison is made between current-mode


adaptive on-time control circuit with and without the ramp generator to understand
the advantage and superiority of the elimination of the subharmonic issue.
114 4 Review of the Adaptive On-time Control …

Fig. 4.10 Circuit diagram of VCC


ramp generator
VCC
R1
G1
VA

R2 VRAMP
VCCS

CLK Q C

The operation conditions are 395 kHz of switching frequency, 5 V of input voltage,
0.75 V of reference voltage, 3.3 V of output voltage, and 10 A of output load
current. The output ceramic capacitor is 44 lF, and its ESR is 1 mX.
Figure 4.11a shows the simulation results at the steady status for the
current-mode adaptive on-time control circuit without a ramp generator. The
red-colored waveform of the output voltage signal is approximately 100 mV with a
subharmonic issue, and the blue-colored waveform of the VUG signal is a PWM
signal. The VUG signal exhibits a subharmonic issue, so the width of the VUG signal
is not the same and the switching frequency cannot be measured, unlike when the
VUG signal does not exhibit a subharmonic issue. In addition, the current-mode
adaptive on-time control circuit without a ramp generator has a subharmonic issue
when the duty cycle is larger than 50%. This subharmonic issue does not result in
feedback voltage VFB without the virtual inductor current ripple causing time-delay
effects in the loop.
Figure 4.11b shows the simulation results at the steady status for the current-
mode adaptive on-time control circuit with a ramp generator. The red-colored
waveform of the output voltage signal is approximately 4.6 mV without a sub-
harmonic issue, and the blue-colored waveform of the VUG signal is a PWM signal.
The VUG signal does not exhibit a subharmonic issue, so the width of the VUG signal
is the same, and the switching frequency can be measured at 395.25 kHz.
The current-mode adaptive on-time control circuit has an error amplifier, and the
output terminal of the error amplifier can amplify a voltage difference between VFB
and VREF. Thus, it can be used to avoid subharmonic oscillation and time-delay
effects with double pulses or pulses even when the output capacitor uses ceramic
capacitors. The ramp generator is added to voltage VCS when the duty cycle is larger
than 50%. The current-mode adaptive on-time control circuit with a ramp generator
does not have a subharmonic issue in the loop.
The current-mode adaptive on-time control circuit is highly similar to the
current-mode control circuit for buck converter, but the on-time generator of the
current-mode adaptive on-time control circuit is significantly different from that of
4.3 Current-Mode Adaptive On-time Control Circuit … 115

Fig. 4.11 a Simulation results at the steady status for current-mode adaptive on-time control
circuit without ramp generator. b Simulation results at the steady status for current-mode adaptive
on-time control circuit with ramp generator
116 4 Review of the Adaptive On-time Control …

the current-mode control circuit [36]. The droop can be easily determined based on
this difference. The current-mode adaptive on-time control circuit can generate
multiple pulses at a variable frequency, and the current-mode control circuit can
simply generate a single pulse at a constant frequency when the output load current
is changed.
Figure 4.12 shows the simulation results, which can be used to compare the
current-mode adaptive on-time control circuit with the current-mode control circuit
at the droop [36]. These simulation results use the same operation conditions to
compare the current-mode adaptive on-time control circuit with the current-mode
control circuit, as shown in Table 4.1 [36]. The current-mode adaptive on-time
control circuit can generate multiple pulses with a minimum off-time mechanism to
prevent VOUT from decreasing significantly. The current-mode adaptive on-time
control circuit is useful for the reduction of VOUT peak-to-peak voltage at the
droop. However, the current-mode control circuit can simply increase its PWM
pulse width at the same switching frequency at the droop. Comparison between the
current-mode adaptive on-time control circuit and the current-mode control circuit
at the droop shows that the former can generate more PWM pulses than the latter.
Therefore, the current-mode adaptive on-time control circuit can achieve a faster
transient response than the current-mode control circuit.

IOUT
(A)

VUG
(V)

VUG
(V)

VOUT
(V)

t (us)

Fig. 4.12 Compare between current-mode adaptive on-time control circuit and current-mode
control circuit at the droop
4.4 Adaptive On-time Control Circuits with Adaptive … 117

Table 4.1 Operation conditions for current-mode adaptive on-time control and current-mode
control circuit for buck converter
VIN 19 V FS 320 kHz L 2.2 lH
VOUT 1.05 V RD1 8 kX COUT 330 lF*2
IOUT 1*9 A (8 ls) RD2 20 kX RCO 5 mX
VREF 0.75 V RC 1.5 MX CC 40 pF

4.4 Adaptive On-time Control Circuits with Adaptive


Voltage Positioning Design for Voltage Regulators

The AVP design of the adaptive on-time control circuit for VRs is as shown in
Fig. 4.13 [3, 14–17, 37–39]. The EA is an error amplifier and VCOMP signal is the
output of error amplifier. Two input signals of error amplifier are the feedback
signal VFB and the VVID voltage, if the user wants to change the output voltage, and
the user should be based on VR12 VID code table setting VID bits by I2C interface
[40, 41]. S11, S12 and S13 are the upper-side switches, S21, S22 and S23 are the
lower-side switches, L1, L2, and L3 are the output inductors, and RCO is the ESR of
the output capacitor CO. The current source ILOAD is the output load current. The
feedback resistors, R1 and R2 can be designed by load line requirement, so the
output voltage VOUT is not equal to the VVID voltage. The output voltage VOUT
depends on the output load current, if the system is operated from the light load to
the heavy load and the output voltage VOUT also should be reduced to meet the load
line requirement. The trigger signal VTRIG is the output of comparator to control
on-time generator.
The steady state waveforms of the AVP design are as shown in Fig. 4.14. The
voltage at the output terminal of the error amplifier is VCOMP. The VOUT is larger than
VOUT_TARGET in steady state and hence larger the VOUT back to its expected value.
The VOUT_OFS voltage is a droop between the VOUT voltage and the VOUT_TARGET
voltage. The AVP design of the adaptive on-time control circuit for VRs has a DC
offset of the VOUT voltage, however, the VRs should be designed a high accuracy the
VOUT voltage, otherwise this offset of the VOUT voltage may impact the load line
specification. Refer to the steady-state control signals at the “A” point shown in
Fig. 4.14, and Eqs. (4.10)–(4.16) are obtained [3, 14–17, 37–39].

VCS;Vally ¼ VCOMP;Peak ð4:10Þ

DIL
VVID þ Ri  ðIL  Þ ¼ VCOMP ð4:11Þ
2
DIL DVOUT
VVID þ Ri  ðIL  Þ ¼ VVID þ AV  ðVVID  ðVOUT  ÞÞ ð4:12Þ
2 2
118 4 Review of the Adaptive On-time Control …

S13 L3

VUG3 IL3
VLG3 S23
Driver3

S12 L2

VUG2 IL2
VLG2 S22
Driver2

VIN S11 L1 VOUT


VUG1 IL1 ILOAD
VLG1 S21 RCO
Driver1
CO
R1 C1
VVID VFB
Interleaved Ri
and PSM

+
-
Control EA R2 C2

VCS
On-Time VTRIG
-

Generator + VCOMP

VIN VOUT

Fig. 4.13 AVP design of the adaptive on-time control circuit for VRs

DV OUT DI L
AV  V OUT ¼ AV  ðV VID þ Þ  Ri  ðI L  Þ ð4:13Þ
2 2
DVOUT Ri DIL
VOUT ¼ ðVVID þ Þ  ðIL  Þ ð4:14Þ
2 AV 2

Ri
RDROOP ¼ ð4:15Þ
AV

R2
AV ¼ ð4:16Þ
R1

AV is the desired error amplifier gain. Ri is the internal current sense amplifier
gain. RDROOP is the current sense resistor. This control also implements the AVP
4.4 Adaptive On-time Control Circuits with Adaptive … 119

VCS
VCOMP A

t
VOUT_OFS
VOUT
VOUT_TARGET

t
VUGx

t
VTRIG

Fig. 4.14 Steady state waveforms of the AVP design

function easily. RDROOP should be designed to determine the load line. It is the
equivalent load line resistance as well as the desired static output impedance.
An optimized compensation of a multiphase voltage regulator allows for the best
possible load step response of its output. A type-II compensator with one pole and
one zero is adequate for proper compensation. A prior design procedure shows how
to decide the resistive feedback components of an error amplifier gain. C1 and C2
can be calculated for compensation by Eq. (4.17) [40, 41]. The target is to achieve
constant resistive output impedance over the widest possible frequency range.

Ri 1 þ s=ðR1  C1 Þ
GVC ðs) ¼ ð4:17Þ
RDROOP 1 þ s=ðR2  C2 Þ

4.5 Summary

The conventional COT control circuit is more suitable to be used for fast load
transient power supplies application and it can generate the fixed on-time width to
control the driver circuit and achieve the voltage regulation. The on-time generator
circuit of the conventional COT control circuit can generate the fixed on-time width
to control the driver circuit and achieve the voltage regulation if the conventional
120 4 Review of the Adaptive On-time Control …

COT control circuit wants to regulate a high VOUT and an increase in switching loss
occurs.
The adaptive on-time control circuits sample the VIN signal and VOUT signal to
adjust the on-time width to control the driver circuit and achieve the voltage reg-
ulation. Two types of the adaptive on-time control circuits based on connection
form are typically used, namely, constant frequency on-time control circuit and
constant current ripple on-time control circuit.

References

1. E. Stanford, Power technology roadmap for microprocessor voltage regulators. Presentation at


PSMA, Feb 2003
2. 2001 technology roadmap for semiconductors. Available: http:// www.intel.com/research/
silicon/AlanAllanIEEEComputer0102.pdf
3. K. Yao, Y. Ren, J. Sun, K. Lee, M. Xu, J. Zhou, F.C. Lee, Adaptive voltage position design
for voltage regulators, in Proceedings IEEE Applied Power Electronics Conference and
Exposition conference, pp. 272–278, 2004
4. H. Mao, L. Yao, C. Wang, I. Batarseh, Analysis of Inductor current sharing in non isolated
and isolated Multiphase dc–dc converters. IEEE Trans. Ind. Electron. 54(6), 3379–3388, Dec
2007
5. P.-W. Lee, Y.-S. Lee, D.K.W. Cheng, X.-C. Liu, Steady-state analysis of an interleaved boost
converter with coupled inductors. IEEE Trans. Ind. Electron. 47(4), 787–795 (2000)
6. H.N. Nagaraja, D. Kastha, A. Patra, Design principles of a symmetrically coupled inductor
structure for multiphase synchronous buck converters. IEEE Trans. Ind. Electron. 58(3), 988–
997 (2011)
7. L.-P. Wong, D.K.-W. Cheng, M.H.L. Chow, Y.-S. Lee, Interleaved three-phase forward
converter using integrated transformer. IEEE Trans. Ind. Electron. 52(5), 1246–1260 (2005)
8. J. Abu-Qahouq, H. Mao, I. Batarseh, Multiphase voltage-mode hysteretic controlled dc–dc
converter with novel current sharing. IEEE Trans. Power Electron. 19(6), 1397–1407 (2004)
9. H. Mao, L. Yao, C. Wang, I. Batarseh, Analysis of inductor current sharing in non isolated
and isolated multiphase dc–dc converters. IEEE Trans. Ind. Electron. 54(6), 3379–3388, Dec
2007
10. P.-W. Lee, Y.-S. Lee, D.K.W. Cheng, X.-C. Liu, Steady-state analysis of an interleaved boost
converter with coupled inductors. IEEE Trans. Ind. Electron 47(4), 787–795 (2000)
11. H.N. Nagaraja, D. Kastha, A. Patra, Design principles of a symmetrically coupled inductor
structure for multiphase synchronous buck converters. IEEE Trans. Ind. Electron 58(3), 988–
997 (2011)
12. L.-P. Wong, D.K.-W. Cheng, M.H.L. Chow, Y.-S. Lee, Interleaved three-phase forward
converter using integrated transformer. IEEE Trans. Ind. Electron. 52(5), 1246–1260 (2005)
13. J. Abu-Qahouq, H. Mao, I. Batarseh, Multiphase voltage-mode hysteretic controlled dc–dc
converter with novel current sharing. IEEE Trans. Power Electron. 19(6), 1397–1407 (2004)
14. J.R. Huang, C.H. Wang, C.J. Lee, K.L. Tseng, and D. Chen, Native AVP control method for
constant output impedance of DC power converters, in Proceedings IEEE Power Electronics
Specialists Conference, pp. 2023–2028, 2007
15. A. Waizman, C.Y. Chung, Resonant free power network design using extended adaptive
voltage positioning (EAVP) methodology. IEEE Trans. Adv. Packag. 24, 236–244 (2001)
16. M. Lee, D. Chen, K. Huang, E. Tseng and B. Tai, Compensator design for adaptive voltage
position (AVP) for multiphase VRMs, in Proceedings IEEE Power Electronics Specialists
Conference, 2006
References 121

17. K. Yao, Y. Meng, P. Xu, F.C. Lee, Design considerations for VRM transient response based
on the output impedance, in Proceedings IEEE Applied Power Electronics Conference and
Exposition conference, pp. 14–20, 2002
18. W.W. Chen, J.F. Chen, T.J. Liang, L.C. Wei, J.R. Huang, W.Y. Ting, A novel quick response
of RBCOT with VIC ripple for buck converter. IEEE Trans. Power Electron. 28, 4299–4308
(2013)
19. W.W. Chen, J.F. Chen, T.J. Liang, J.R. Huang, L.C. Wei, W.Y. Ting, Implementing dynamic
quick response with high-frequency feedback control of the deformable constant on-time
control for buck converter on-chip. IET Power Electron 6(4), 383–391 (2013)
20. W.W. Chen, J.F. Chen, T.J. Liang, S.F. Hsiao, J.R. Huang, W.Y. Ting, Improved transient
response using HFFC circuit of the CCRCOT with native AVP design for voltage regulators.
IET Power Electron 6, 1948–1955 (2013)
21. W.W. Chen, J.F. Chen, T.J. Liang, J.R. Huang, and W.Y. Ting, Improved Transient Response
Using HFFC in Current-Mode CFCOT Control for Buck Converter, in Proceedings IEEE
International Conference on Power Electronics and Drive Systems (PEDS), pp. 546–549
22. C.J. Chen, D. Chen, C.W. Tseng, C.T. Tseng, Y.W. Chang, K.C. Wang, A Novel
Ripple-Based Constant On-time Control with Virtual Inductor Current Ripple for Buck
Converter with Ceramic Output Capacitors, in Proceedings IEEE Applied Energy Conversion
Congress and Exposition conference, pp. 1244–1250, 2011
23. R. Redl, J. Sun, Ripple-based control of switching regulators an overview. IEEE Trans. Power
Electron. 24, 2669–2680 (2009)
24. J. Sun, Characterization and performance comparison of ripple-based control for voltage
regulator modules. IEEE Trans. Power Electron. 21, 346–353 (2006)
25. W. Huang, A new control for multi-phase buck converter with fast transient response, in
Proceedings IEEE Applied Power Electronics Conference and Exposition conference,
pp. 273–279, 2001
26. J. Li, F.C. Lee, Modeling of V2 current-mode control, in Proceedings IEEE Applied Power
Electronics Conference and Exposition conference, pp. 298–304, 2009
27. K.D.T. Ngo, S.K. Mishra, M. Walters, Synthetic-ripple modulator for synchronous buck
converter. Proc. IEEE Power Electron. 3, 148–151 (2005)
28. Y.H. Lee, S.J. Wang, K.H. Chen, Quadratic differential and integration technique in V2
control buck converter with small ESR capacitor. Proc. IEEE Trans. Power Electron. 25, 829–
838 (2010)
29. M.Y. Yen, P. Mok, A constant frequency output-ripple voltage-based buck converter without
using large ESR capacitor. IEEE Trans. Circuits Syst. 55, 748–752 (2008)
30. K.Y. Cheng, F. Yu, P. Mattavelli, F.C. Lee, Characterization and performance comparison of
digital V2-type constant on-time control for buck converters, IEEE Control and Modeling for
Power Electronics conference, pp. 1–6, June 2010
31. J. Li, Current-mode control: modeling and its digital application Ph.D. thesis, Virginia
Polytechnic Institute and State University, 2009
32. J. Li and F.C. Lee, New modeling approach and equivalent circuit representation for
current-mode control. IEEE Trans. Power Electron. 1218–1230, May 2010
33. S.J. Wang, Y.H. Lee, Y.C. Lai, K.H. Chen, Quadratic differential and integration technique in
V2 control buck converter with small ESR capacitor, in Proceedings IEEE Custom Integrated
Circuits Conference, pp. 211–214, 2009
34. R. Redl, G. Reizik, Switched noise filter for the buck converter using the output ripple as the
PWM ramp, in Proceedings IEEE Appl. Power Electron. Conference, pp. 918–924, 2005
35. R. Redl and T. Schiff, A new family of enhanced ripple regulators for power-management
applications, in Proceedings of International Exhibition and Conference Eur., Nuremburg,
Germany, pp. 255–268, 2008
36. W.W. Chen, J.F. Chen, T.J. Liang, Dynamic ramp control in current-mode adaptive on-time
control for buck converter on chip, in Proceedings of IEEE Future Energy Electronics
Conference and ECCE Asia (IFEEC 2017–ECCE Asia), pp. 280–285, 2017
122 4 Review of the Adaptive On-time Control …

37. P.L. Wong, Performance improvements of multi-channel interleaving voltage regulator


modules with integrated coupling inductors, Dissertation of Virginia Polytechnic Institute and
State University, March 2001
38. S.K. Mishra, Design-oriented analysis of modern active droop controlled power supplies.
IEEE Trans. Ind. Electron. 56(9), 3704–3708 (2009)
39. J.A.A. Qahouq, V. Arikatla, Power converter with digital sensorless adaptive voltage
positioning control scheme. IEEE Trans. Ind. Electron. 58(9), 4105–4116 (2010)
40. Richtek Tech. Corp., Dual output 3-Phase + 2-Phase PWM controller for CPU and GPU Core
power supply, RT8885A Datasheet, 2012
41. Richtek Tech. Corp., Multi-Phase PWM controller for CPU core power supply, RT8859 M
Datasheet, 2014
Chapter 5
Adaptive On-time Control Circuit
for Buck Converters

5.1 Increasing Light Load Efficiency with PSM Mode

Today’s products, particularly those designed to reduce standby power loss and
increase system loop response, feature high performance and low loss because of
their system design. The conventional fixed-frequency PWM control scheme for
power converters [1–13] is commonly used with current-mode control instead of
voltage-mode control. However, adaptive on-time control circuits [14–25] can
achieve a faster transient response than the PWM control scheme because the
former uses a comparator to control the on-time generator without an error
amplifier, as shown in Fig. 5.1. Thus, adaptive on-time control circuits do not
exhibit system loop delay from the error amplifier. However, adaptive on-time
control circuits do not meet the requirements of output equipment or devices to
achieve a faster transient response. This chapter shows that adaptive on-time control
circuits with a quick dynamic response can achieve a faster transient response than
those without a quick dynamic response for buck converters.
In Fig. 5.1, SA and SB are the switches, L is the output inductor, and RCO is the
ESR of the output capacitor CO. The current source IOUT is the output load current,
and RD1 and RD2 are the feedback resistors to determine the output voltage. VFB is a
feedback voltage. The reference voltage VREF is created inside the IC. The on-time
generator circuit samples the VIN signal and VOUT signal to adjust the on-time width
to control the driver circuit and achieve the voltage regulation.
The ZCD function [26–30] is widely used to maintain the inductor current at
zero under light load for buck converter ICs. Thus, the ZCD function can prevent
reverse current from flowing back to the source to improve conversion efficiency.
Under light load, the ZCD function for each switching period can be divided into
three operation modes. In mode 1, the SA switch is turned ON, and buck converter
begins to charge the output capacitors and delivers energy to output equipment or
devices. In mode 2, the SA switch is turned OFF immediately after the SB switch is
turned ON to discharge from the output capacitors and deliver energy to output

© Springer Nature Singapore Pte Ltd. 2018 123


W.-W. Chen and J.-F. Chen, Control Techniques for Power
Converters with Integrated Circuit, Power Systems,
https://doi.org/10.1007/978-981-10-7004-4_5
124 5 Adaptive On-time Control Circuit for Buck Converters

VIN SA L VOUT
VLX

VUG IL IOUT
VLG RCO
SB
CO
RD1
Driver

VFB
On-Time VTRIG
Generator
VREF RD2

VIN VOUT

Fig. 5.1 Adaptive on-time control circuits for buck converter

equipment or devices. In mode 3, the inductor current is lower than or equal to zero,
and the SB switch is turned OFF. The inductor current flows through the body diode
of the SB switch until the next duty cycle is needed to drive the SA switch. Thus,
switching loss can be reduced significantly, as shown in Fig. 5.2. Figure 5.2 shows
the VLX voltage and the IL current. VLX can obtain the control signal information of

VLX (5V/2.5µs)

IL (500mA/2.5µs)

Mode 3
Mode 1 Mode 2

Fig. 5.2 Operation with ZCD function at the light load for buck converter
5.1 Increasing Light Load Efficiency with PSM Mode 125

VUG and VLG voltages. When the VLX voltage is at the maximum level, it is equal to
the input voltage because the SA switch is turned ON at this status. When the VLX
voltage is at a negative level, it is equal to the voltage drop of RDS for the SB switch
because the SB switch is turned ON at this status. The SA and SB switches are turned
OFF when the IL current is equal to zero, and the VLX voltage is equal to the VOUT
voltage. To save cost, most IC products include SA and SB switches on chip, so the
user can measure the VLX voltage to obtain VUG and VLG voltages.
Pulse-skipping modulation (PSM) mode [26–30] is also widely used in ICs to
increase the efficiency under light load for buck converters. The PSM mode can
reduce the number of control pulses under light loads. The conventional
fixed-frequency PWM control scheme needs an extra circuit to skip the original
control pulses to operate at PSM mode. However, adaptive on-time control circuits
can directly achieve a native PSM mode at a light load because the control pulses of
these circuits depend on the output loading condition. Thus, the number of control
pulses at a light load is less than that at a heavy load. The usual PSM mode of
combining adaptive on-time control circuits with the ZCD function can best reduce
the switching loss.
To further understand the advantage and superiority of this PSM mode, the plans
based the same operation conditions to compare the adaptive on-time control cir-
cuits with PSM mode and the current-mode control circuit at the light load with the
ceramic output capacitors 44 µF.
The conditions are as follows:
(1) Input voltage (VIN): 12 V
(2) Output voltage (VOUT): 1.05 V
(3) Output load current (IOUT): 10 mA
(4) Switching frequency (FS): 630 kHz
(5) MOSFET (SA, SB): BSC0909NS * 3
(6) Feedback resistors (RD1, RD2): 13.2 kX and 12 kX
(7) Main inductor (L): IHLP4040DZER1R0MA1 (1 lH)
(8) Output capacitors (CO): 22 lF/6.3 V (RCO: 3 mX) * 2
(9) Reference voltage (VREF): 0.5 V.
Figure 5.3 compares the experimental results of the adaptive on-time control
circuits with PSM mode and current-mode control circuit at the light load.
Figure 5.3a represents the VLX voltage and the IL current for current-mode control
circuit at the light load. Figure 5.3b represents the VLX voltage and the IL current for
adaptive on-time control circuits with PSM mode at the light load. The signals are
measured based on the same output load current. The switching frequency of the
current-mode control circuit maintains 630 kHz. The switching frequency of the
adaptive on-time control circuits with PSM mode is based on the output loading
condition and the switching frequency is measured 33.3 kHz at output current
10 mA. The number of control pulses for current-mode control circuit is larger than
adaptive on-time control circuits with PSM mode, so the current-mode control
circuit has a large switching loss. In addition, the IL current of the adaptive on-time
126 5 Adaptive On-time Control Circuit for Buck Converters

VLX (10V/4µs)

IL (500mA/4µs)

(a) Current-mode control circuit

VLX (10V/4µs)

IL (500mA/4µs)

(b) Adaptive on-time control circuits with PSM mode

Fig. 5.3 Comparison of the experimental results between adaptive on-time control circuits with
PSM mode and current-mode control circuit at the light load

control circuits with PSM mode is maintained zero current during 28 ls, so it is
useful to save conduction loss on the RDS for the SB switch, because the inductor
current is through the body diode of the SB switch. Based on these reasons, the
5.1 Increasing Light Load Efficiency with PSM Mode 127

adaptive on-time control circuits with PSM mode can significantly achieve a better
efficiency at the light load.
Figure 5.4 shows one control pulse of the experimental results of adaptive
on-time control circuits with PSM mode and the current-mode control circuit at a

VLX (10V/80ns)

145ns

IL (500mA/80ns)

(a) Current-mode control circuit

VLX (10V/80ns)

145ns

IL (500mA/80ns)

(b) Adaptive on-time control circuits with PSM mode

Fig. 5.4 One control pulse of the experimental results between adaptive on-time control circuits
with PSM mode and current-mode control circuit at the light load
128 5 Adaptive On-time Control Circuit for Buck Converters

Fig. 5.5 Comparison of the 100


(2)
efficiency results between 90
adaptive on-time control 80
circuits with PSM mode and 70

Efficiency (%)
current-mode control circuit (1)
60
50
40
30
20
(1) Current - Mode Control
10 (2) AOT Control with PSM
0
0.001 0.01 0.1 1 10
Output Current (A)

light load. The signals are measured based on the same output load current. The
width of the control pulse of the current-mode control circuit is 145 ns, and it is
equal to the width of the control pulse of the adaptive on-time control circuits with
PSM mode. The adaptive on-time control circuits can sample the VIN and VOUT
voltages to control the on-time, similar to the duty cycle of the current-mode control
circuit. Even when the VIN and VOUT voltages are changed, the system still
maintains the same duty cycle. The same width of the control pulse can generate a
similar rise time of the IL current. In addition, the peak-to-peak IL current of the
current-mode control circuit is 1 A, and it is equal to the width of the control pulse
of the adaptive on-time control circuits with PSM mode.
Figure 5.5 compares the efficiency results between adaptive on-time control
circuits with PSM mode and current-mode control circuit. Adaptive on-time control
circuits with PSM mode at the light load results in lower switching frequency and
lower switching loss compared with current-mode control circuit. When the output
load current is <0.2 A, the efficiency with adaptive on-time control circuits with
PSM mode is significantly improved. Because the current-mode control circuit at
the light load, it still allows this converter to remain at a fixed frequency, and the
duty cycle becomes the input voltage and the output voltage of the decision. This
fixed frequency results in a large switching loss, and the converter can no longer
achieve a low loss. Minimizing the standby power loss is usually preferred to
reducing switching loss.

5.2 On-time Generator Circuit of the Adaptive On-time


Control Circuits for Buck Converters

Two types on-time generator circuit of the adaptive on-time control circuits based
on connection form are typically used, namely, constant frequency on-time control
circuit and constant current ripple on-time control circuit [1, 31–34].
5.2 On-time Generator Circuit of the Adaptive On-time Control Circuits … 129

VDD

VIN R1 TON
G1
On-Time

VTRIG VOUT
S1 C1

Fig. 5.6 On-time generator circuit of the constant frequency on-time control circuit

On-time generator circuit of the constant frequency on-time control circuit is as


shown in Fig. 5.6. The TON pin is inside the IC, and the resistor R1 can be placed
between TON pin and input voltage to determine the on-time width. VOUT needs to
be monitored to limit the voltage from capacitor C1 by Eq. (5.1). VOUT divided by
VIN is the duty cycle, and R1, C1, and G1 are constant values, like the switching
period TS, as shown in Eq. (5.2). G1 is the gain of the current-controlled current
source. Even if the input voltage VIN and output voltage VOUT are changed, the
system still maintains the same switching frequency [1, 31–34].

VOUT R1  C1
TON ¼  ð5:1Þ
VIN G1

VOUT
TON ¼  TS ð5:2Þ
VIN

Constant frequency on-time control circuit is different from the conventional


COT control circuit because it prevents the generation of the same on-time width at
a high VOUT. Regardless of changes in VIN or VOUT, the conventional COT control
circuit still generates the same on-time width. If the conventional COT control
circuit wants to regulate a high VOUT and needs to generate more on-time pulses, an
increase in switching loss occurs. Thus, constant frequency on-time control circuit
is suitable for a wide range output voltage.
Experimental results of the switching frequency versus the input voltage for
constant frequency on-time control circuit are as shown in Fig. 5.7. The operation
conditions are set at 3.3 V output voltage and 0.1 A output current to measure the
switching frequency for 4–21 V input voltage. When R1 is equal to 250 kX, con-
necting VIN and TON pin, the switching frequency is 267 kHz, regardless of
changes in VIN. If R1 is reduced to 150 kX, the switching frequency changes from
267 to 444 kHz, because R1 directly affects the on-time width. R1 and the switching
frequency are inversely related.
130 5 Adaptive On-time Control Circuit for Buck Converters

Switching Frequency vs. Input Voltage


Output Voltage=3.3V, Output Current=0.1A
500
Switching Frequency (kHz) (1)
450

400

350

300 (2)
250
(2) R1=250 kΩ
200 (1) R1=150 kΩ
150
2 4 6 8 10 12 14 16 18 20 22
Input Voltage (V)

Fig. 5.7 Experimental results of the switching frequency versus the input voltage for constant
frequency on-time control circuit

Inductor Current Ripple vs. Input Voltage


Output Voltage=3.3V, Output Current=0.1A
4.5
Inductor Current Ripple (A)

4
(2)
3.5
3
2.5
2 (1)

1.5
1 (2) R1=250 kΩ
(1) R1=150 kΩ
0.5
2 4 6 8 10 12 14 16 18 20 22
Input Voltage (V)

Fig. 5.8 Experimental results of the inductor current ripple versus the input voltage for constant
frequency on-time control circuit

Experimental results of the inductor current ripple versus the input voltage for
constant frequency on-time control circuit are as shown in Fig. 5.8. The operating
conditions are set 3.3 V output voltage and 0.1 A output current to measure the
inductor current ripple for 4–21 V input voltage. If the inductance is invariant, then
the inductor current ripple is related to the voltage difference between the input
voltage and output voltage and is also dependent on the on-time TON, as shown by
Eq. (5.3). By substituting Eq. (5.2) into Eq. (5.3), the inductor current ripple is
5.2 On-time Generator Circuit of the Adaptive On-time Control Circuits … 131

obtained as Eq. (5.4). The inductor current ripple increases as the input voltage is
increased. Hence, it does not maintain a constant value. When R1 is increased from
150 to 250 kX, the on-time width is also increased, resulting in a larger inductor
current ripple.

VIN  VOUT
DIL ¼  TON ð5:3Þ
L
VOUT ð1  ðVOUT =VIN ÞÞ
DIL ¼  TS ð5:4Þ
L

On-time generator circuit of the constant current ripple on-time control circuit is
as shown in Fig. 5.9. The input voltage and output voltage are sampled to deter-
mine the on-time width. The TON is an IC pin connected to VIN through a resistor.
This circuit also samples VOUT by Eq. (5.5). The voltage drop between the input
voltage and output voltage depends on the inductor current ripple, as shown by
Eq. (5.6). Therefore, even if the voltage drop between the input voltage and output
voltage is changed, the inductor current ripple maintains a constant value. Hence,
the constant current ripple on-time control circuit is suitable for applications with
different the input voltage and output voltage, requiring the same output voltage
ripple, such as CPUs [1, 31–34].

R1
TON ¼  C1  VREF1 ð5:5Þ
VIN  VOUT

L
TON ¼  DIL ð5:6Þ
VIN  VOUT

Experimental results of the switching frequency versus the input voltage for
constant current ripple on-time control circuit are as shown in Fig. 5.10. The
operation conditions are set at 3.3 V output voltage and 0.1 A output current to

VDD

VIN R1 TON
G1
On-Time

VOUT
VTRIG S1 C1 VREF1

Fig. 5.9 On-time generator circuit of the constant current ripple on-time control circuit
132 5 Adaptive On-time Control Circuit for Buck Converters

Switching Frequency vs. Input Voltage


Output Voltage=3.3V, Output Current=0.1A
500

Switching Frequency (kHz)


(1)
450

400

350

300 (2)
250
(2) R1=250 kΩ
200
(1) R1=150 kΩ
150
2 4 6 8 10 12 14 16 18 20 22
Input Voltage (V)

Fig. 5.10 Experimental results of the switching frequency versus the input voltage for constant
current ripple on-time control circuit

measure the switching frequency for 4–21 V input voltage. When R1 is equal to
150 kX, connecting VIN and TON pin, the switching frequency increases from 296
to 489 kHz when the input voltage is increased from 4 to 21 V. Increasing R1 from
150 to 250 kX, the switching frequency increases from 188 to 322 kHz when the
input voltage is increased from 4 to 21 V.
Experimental results of the inductor current ripple versus the input voltage for
constant current ripple on-time control circuit are as shown in Fig. 5.11. The
operation conditions are set at 3.3 V output voltage and 0.1 A output current to
measure the inductor current ripple for 4–21 V input voltage. If the inductance is

Inductor Current Ripple vs. Input Voltage


Output Voltage=3.3V, Output Current=0.1A
3.5
Inductor Current Ripple (A)

(2)
3

2.5

2 (1)

1.5

(2) R1=250 kΩ
1
(1) R1=150 kΩ
0.5
2 4 6 8 10 12 14 16 18 20 22
Input Voltage (V)

Fig. 5.11 Experimental results of the inductor current ripple versus the input voltage for constant
current ripple on-time control circuit
5.2 On-time Generator Circuit of the Adaptive On-time Control Circuits … 133

invariant, then the inductor current ripple is related to the voltage drop between the
input voltage and output voltage and is also dependent on the on-time TON. The
inductor current ripple maintains a consistent value. When R1 is increased from 150
to 250 kX, the inductor current ripple also increases from 1.8 to 3.0 A.

5.3 Comparison of Quick Dynamic Response


and Conventional Quick Response of the On-time
Generator Circuit

Two types of improved transient response circuits are typically used, namely,
external setting by component or voltage source and internal fixed by IC design.
Figure 5.12 shows the external setting quick response of the on-time generator
circuit for constant frequency on-time control circuit. The operation principle of the
external setting quick response requires the addition of a voltage source VQRSET to
determine the QR-time width. VQRSET is connected to the QRSET pin. If the
QR-time needs a long QR-time width, VQRSET should be larger than VOUT [1, 31–
34].
V′OUT determines the peak voltage of C1. Thus, if V′OUT is high, the peak voltage
increases, causing a long on-time width. V′OUT is implemented to provide a
ripple-less voltage because it can keep the switching frequency constant without

VIN R1 TON G1
VDD
On-Time

VTRIG S1 C1 V'OUT

G2
VDD
QR-Time

VQRTH
VOUT
C2
S2
QRSET
RLOW

QR trigger VQRSET
CLOW
circuit

Fig. 5.12 External setting quick response of the on-time generator circuit for constant frequency
on-time control circuit
134 5 Adaptive On-time Control Circuit for Buck Converters

under output voltage ripple and noise interference. V′OUT is implemented to sample
the Phase signal through a second-order low-pass filter, producing a similar VOUT
signal. Therefore, V′OUT is ripple-less and has slower transient response than the
VOUT signal.
The QR trigger circuit samples VOUT and uses the low-pass filter to generate
VOUT delay. This trigger circuit also needs the voltage signal VQRTH. In the
steady-state operation, the VOUT drop cannot trigger the QR-time. When this abrupt
voltage drop is lower than VQRTH, the output of QR trigger circuit changes from a
high-level signal to a low-level signal to turn OFF the S2 switch and then the
on-time generator circuit generates the QR-time width. The low-pass filter fre-
quency for RLOW and CLOW should be designed much smaller than the switching
frequency which can avoid this system to fail in its operation with QR-time. VQRTH
can be designed in the IC or set by the user through the QRTH pin. The main
advantage of the external setting quick response is that it can depend on the load
conditions to design the VQRSET value, generating a suitable QR-time width in this
system. However, this method requires two extra IC pins to achieve the quick
response function [1, 31–34].
Internal fixed quick response of the on-time generator circuit for constant current
ripple on-time control circuit is as shown in Fig. 5.13. This circuit has two on-time
generator circuits to generate the on-time width and QR-time width. When the
constant current ripple on-time control circuit is operated at steady state, the system
generates the on-time control signal to drive the switches, thereby regulating the
output voltage. The output load current is changed from a light load to a heavy load,
which makes the output voltage dropped significantly thereby causing the system
to generate the QR-time control signal and achieve the quick response function.

VIN R1 TON G1
VDD
On-Time

VTRIG S1 C1 VREF1

G1
VDD
QR-Time

VOUT VTRIG C2
S2 VREF1

Fig. 5.13 Internal fixed quick response of the on-time generator circuit for constant current ripple
on-time control circuit
5.3 Comparison of Quick Dynamic Response and Conventional Quick … 135

The QR-time width must be longer than the on-time width. The implementation of
the QR-time generator circuit is similar to that of the on-time generator circuit. The
only difference is that C2 must be greater than C1 [1, 31–34].
Implementing the internal fixed quick response can also improve the transient
response for constant current ripple on-time control circuit. The main advantage of
this circuit is that it does not need an extra IC pin to achieve the quick response
function. However, it has some disadvantages: (1) The IC cannot adjust the
QR-time width. Hence, when the output load current is changed from a light load to
a heavy load, the system still outputs the same width of the QR-time control signal;
(2) Designing a hysteresis level to generate the QR-time control signal is difficult.
Hence, when the output load current is different, this circuit may cause the system
to fail in its operation with QR-time control signal.
Quick dynamic response of the on-time generator circuit for constant frequency
on-time control circuit is proposed in this chapter, as shown in Fig. 5.14. This quick
dynamic response does not require an extra pin to achieve the quick response
function of the IC. This quick dynamic response consists of a resistor Rq and a
capacitor Cq connected in series between VOUT and TON pin. It operates under the
principle that the high-pass filter allows only high frequency signal to pass from
VOUT to the TON pin, as shown in Eq. (5.7). However, the frequency FRC of the
high-pass filter must be larger than or equal to the switching frequency FS, so that
the steady state operation of the system is not affected. Finally, the user may design
Rq and Cq based on the worst-case operation (maximum output loading step)
[1, 31–34].

VIN

R1 I1

TON G1
VDD
On-Time

Rq

I2 VTRIG C1
S1 V'OUT
Cq

VOUT

Fig. 5.14 Quick dynamic response of the on-time generator circuit for constant frequency on-time
control circuit
136 5 Adaptive On-time Control Circuit for Buck Converters

1
FRC ¼  FS ð5:7Þ
2p  Rq  Cq

V′OUT determines the peak voltage of C1. Thus, if V′OUT is high, the peak voltage
increases, causing a long on-time width. V′OUT is implemented to provide a
ripple-less voltage because it can keep the switching frequency constant without
under output voltage ripple and noise interference. V′OUT is implemented to sample
the Phase signal through a second-order low-pass filter producing a similar VOUT
signal. Therefore, V′OUT is ripple-less and has a slower transient response than
VOUT.
When a light load quickly transforms to a heavy load, VOUT momentarily drops
through the coupling by Cq, which inducing Rq to cause a voltage drop. The voltage
drop of Rq will form a current of I2. Thus, Rq needs to be designed first because I2
can directly change the on-time width as shown in Fig. 5.15.
However, a longer on-time width induces the converter to deliver more energy
from the input terminal to the output loading. If Rq is reduced, the on-time width
becomes longer. Cq should be designed with the frequency of high-pass filter. The
calculation of the Laplace transform formula is shown in Eq. (5.8).

C1 0
TON ðs) ¼    VOUT ð5:8Þ
G1 ððVIN Þ=ðR1 ÞÞ þ sCq VOUT = 1 þ sCq Rq

The advantages of this quick dynamic response of the on-time generator circuit
are as follows: (1) It clearly generates a longer on-time width that is proportional to
the output voltage drop; (2) Adaption to the on-time width depends on the load
conditions of the system; (3) It only uses one on-time generator circuit; thus, it is
very convenient to design and apply; (4) It does not require an extra pin to achieve
the quick response function.

Fig. 5.15 Control signals for VOUT


quick dynamic response of [(1+RD1/RD2)*VREF]
the on-time generator circuit
for constant frequency
on-time control circuit I2
(0)

I1+I2
(I1)

IOUT (Heavy Load)

(Light Load)
On-Time
(VDD)

(0)
5.4 Experimental Results 137

5.4 Experimental Results

The verification of the experimental and simulation results is conducted to prove the
feasibility and performance with quick dynamic response of the constant frequency
on-time control circuit for Buck converter.
The specifications are as follows:
(1)Input voltage (VIN): 12 V
(2)Output voltage (VOUT): 3.3 V
(3)Output load current (IOUT): 18 A @ 1.2857 A/ls
(4)Switching frequency (FS): 267 kHz
(5)MOSFET (SA, SB): BSC0909NS * 3
(6)Feedback resistors (RD1, RD2): 47 and 15 kX
(7)Main inductor (L): IHLP4040DZER1R0MA1 (1 lH)
(8)Output capacitors (CO): 330lF/6.3 V (RCO:4.5 mX) * 1
(9)Reference voltage (VREF): 0.8 V
(10)Quick dynamic response: G1 = 1(A/A); R1 = 250 kX; C1 = 15 pF;
Rq = 500 X; Cq = 1.2 nF; FRC = 262 kHz
11) External setting quick response: RLOW = 600 kX; CLOW = 2 pF;
VQRSET = 4.125 V; VQRTH = 100 mV
Figure 5.16 shows the chip layout of the control IC. The driver circuit and TON
control are both occupied, marking an area with a die size of 880 lm  880 lm.
Evaluation board of the constant frequency on-time control circuit with quick
dynamic response for Buck converter is as shown in Fig. 5.17.
SIMPLIS simulation results at the load transient with quick dynamic response
(see Fig. 5.14) and without quick response (see Fig. 5.6) are as shown in Fig. 5.18.
Without quick response cannot directly change its on-time width at the load tran-
sient. The blue-colored waveform represents the output voltage signal without
quick response, and the red-colored waveform represents the output voltage signal
with quick dynamic response. These output voltage signals are simulated based on
the same output load current as the black-colored waveform.
VOUT peak to peak value at the load transient with quick dynamic response is
288 mV and VOUT peak to peak value at load transient without quick response is
354 mV. Thus, VOUT peak to peak value with quick dynamic response is lower by
66 mV than without quick response. In addition, the settling time with quick
dynamic response is shorter than without quick response.
Experimental results at the load transient with quick dynamic response (see
Fig. 5.14) and without quick response (see Fig. 5.6) are as shown in Fig. 5.19.
Without quick response cannot directly change its on-time width at the load tran-
sient. The blue-colored waveform represents the output voltage signal without
quick response, and the red-colored waveform represents the output voltage signal
with quick dynamic response. These output voltage signals are measured based on
the same output load current as the black-colored waveform. The output load
transient is generated by a function generator to generate fast slew rate control
138 5 Adaptive On-time Control Circuit for Buck Converters

Protection
Reference
Driver Circuit

880 µm
ZCD

TON

Fuse

880 µm

Fig. 5.16 Chip layout of the control IC

Fig. 5.17 Evaluation board of the constant frequency on-time control circuit with quick dynamic
response for buck converter
5.4 Experimental Results 139

233mV
VOUT(50mV/20us)

Dynamic Quick Response with HFFC

300mV
VOUT(50mV/20us)

Without Quick Response

IOUT(2A/20us)

Fig. 5.18 Comparison of the simulation results at the load transient with quick dynamic response
and without quick response

Fig. 5.19 Comparison of the


experimental results at the VOUT(100mV/40µs)
260mV
load transient with quick
(1)
dynamic response and
without quick response
330mV
VOUT(100mV/40µs)
(2)

IOUT(10A/40µs)

(3)
(2) Without Quick Response
(1) Dynamic Quick Response with HFFC

signal and drive the MOSFET switch. This implemented method produces faster
load transient than the electronics load. VOUT peak to peak value at the load tran-
sient with quick dynamic response is 260 mV, and VOUT peak to peak value at the
load transient without quick response is 340 mV. Thus, VOUT peak to peak value
variation with quick dynamic response is lower 80 mV than without quick
140 5 Adaptive On-time Control Circuit for Buck Converters

Fig. 5.20 Comparison of the


VUG(20V/4µs)
experimental results at the
droop with quick dynamic (1)
response and without quick
response VUG(20V/4µs)

(2)
VOUT(100mV/4µs)
160mV (3)

VOUT(100mV/4µs)
220mV (4)

(2),(4)Without Quick Response


(1),(3) Dynamic Quick Response with HFFC

response. This quick dynamic response is useful to prevent the VOUT from dropping
significantly.
Experimental results at the droop with quick dynamic response and without quick
response are as shown in Fig. 5.20. The blue-colored waveforms represent the
output voltage and VUG signal without quick response and the red-colored wave-
forms represent the output voltage and VUG signal with quick dynamic response. The
signals are measured based on the same output load current. VOUT peak to peak value
at the droop with quick dynamic response is 160 mV, and VOUT peak to peak value
at the droop without quick response is 220 mV. Thus, VOUT peak to peak value with
quick dynamic response is lower than without quick response, with a difference of
about 60 mV. The setting time at the droop with quick dynamic response is also
shorter than without quick response. In addition, the VUG signal with quick dynamic
response is obviously longer than without quick response at the droop, enabling the
input power to deliver more energy to the output terminal.
Experimental results at the load release with quick dynamic response and
without quick response are as shown in Fig. 5.21. The blue-colored waveforms
represent the output voltage and VUG signal without quick response and the
red-colored waveforms represent the output voltage and VUG signal with quick
dynamic response. The signals are measured based on the same output load current.
The overshoot of VOUT peak to peak value at the load release with quick dynamic
response is 80 mV and the overshoot of VOUT peak to peak value at load release
without quick response is 100 mV. In addition, the settling time at the load release
with quick dynamic response is also shorter than without quick response because
VOUT does not suffer a voltage drop after load release.
The plans to base the identical operation conditions by comparing the external
setting quick response for constant frequency on-time control circuit to understand
the advantage and the superiority of this quick dynamic response further.
Figure 5.22 compares the experimental results at the load transient with quick
dynamic response (see Fig. 5.14) and external setting quick response (see
5.4 Experimental Results 141

Fig. 5.21 Comparison of the


VUG(20V/10µs)
experimental results at the
load release with quick (1)
dynamic response and
without quick response VUG(20V/10µs)

(2)

80mV VOUT(100mV/10µs) (3)

90mV
VOUT(100mV/10µs)
(4)

(2),(4) Without Quick Response


(1),(3) Dynamic Quick Response with HFFC

Fig. 5.22 Comparison of the


experimental results at the 260mV VOUT(100mV/40µs)
load transient with quick
(1)
dynamic response and
external setting quick
response VOUT(100mV/40µs) 290mV
(2)

IOUT(10A/40µs)

(3)
(2) External Setting Quick Response
(1) Dynamic Quick Response with HFFC

Fig. 5.12). The green-colored waveform represents the output voltage signal with
external setting quick response, and the red-colored waveform represents the output
voltage signal with quick dynamic response. These output voltage signals are
measured based on the same output load current as the black-colored waveform.
The output load current is used as a function generator to generate fast slew rate
control signal, driving the MOSFET switch to implement the load transient. This
implemented method is faster load transient than the electronics load.
VOUT peak to peak value at the load transient with quick dynamic response is
260 mV and VOUT peak to peak value at the load transient with external setting
quick response is 290 mV. Thus, VOUT peak to peak value with quick dynamic
response and external setting quick response are close. Quick dynamic response and
external setting quick response are useful to prevent the VOUT from dropping sig-
nificantly, but quick dynamic response does not require an extra pin to achieve the
quick response function of IC.
142 5 Adaptive On-time Control Circuit for Buck Converters

Fig. 5.23 Comparison of the


VUG(20V/4µs)
experimental results at the
droop with quick dynamic (1)
response and external setting
quick response VUG(20V/4µs)

(2)
160mV VOUT(100mV/4µs)
(3)

190mV VOUT(100mV/4µs)
(4)

(2),(4) External Setting Quick Response


(1),(3) Dynamic Quick Response with HFFC

Experimental results at the droop with quick dynamic response and external
setting quick response are as shown in Fig. 5.23. The green-colored waveforms
represent the output voltage and VUG signal with external setting quick response,
and the red-colored waveforms represent the output voltage and VUG signal with
quick dynamic response. The signals are measured based on the same output load
current.
VOUT peak to peak value at the droop with quick dynamic response is 160 mV
and VOUT peak to peak value at the droop with external setting quick response is
190 mV. Thus, VOUT peak to peak value with quick dynamic response and external
setting quick response are also close, having a difference of only about 30 mV. In
addition, the VUG signal with quick dynamic response is obviously proportional to
the output voltage drop. The VUG signal with external setting quick response has the
same QR-time width when the QR trigger circuit changes from a high-level signal
to a low-level signal to turn OFF the S2 switch.

5.5 Summary

This chapter proposes of the constant frequency on-time control circuit with quick
dynamic response for Buck converter. The concept uses the high frequency feed-
back control to filters VOUT at the load transient to change the on-time width
dynamically. The input power can deliver more energy to the output terminal to
prevent the VOUT from dropping significantly.
Both experimental and simulation results confirm that the proposed with quick
dynamic response can significantly improve the transient response of the constant
frequency on-time control circuit. Moreover, the proposed quick dynamic response
has a very simple structure and design of the constant frequency on-time control
circuit for Buck converter.
References 143

References

1. C.J. Chen, D. Chen, C. W. Tseng, C. T. Tseng, Y. W. Chang, K.C. Wang, A novel


ripple-based constant on-time control with virtual inductor current ripple for buck converter
with ceramic output capacitors, in Proceedings IEEE Applied Energy Conversion Congress
and Exposition Conferences, (2011), pp. 1244–1250
2. W.H. Ki, Signal flow graph in loop gain analysis of dc-dc PWM CCM switching converters.
IEEE Trans. Circ. Syst. Part I. 6, 644–654 (1998)
3. W.H. Ki, Analysis of subharmonic oscillation of fixed-frequency current-programming switch
mode power converters. IEEE Trans. Circ. Syst. Part I. 45(1), 104–108 (1998)
4. A.D. Schoenfeld, Y. Yu, ASDTIC control and standardized interface circuits applied to buck,
parallel and buck-boost DC-to-DC power converters (NASA, Washington, DC, NASA
Rep. NASA CR-121106, 1973)
5. C.W. Deisch, Switching control method changes power converter into a current source, in
Proceedings of IEEE Power Electronics Specialists Conferences, (1978), pp. 300–306
6. P.L. Hunter, Converter circuit and method having fast responding current balance and
limiting, U.S. Patent 4 002 963, 1 Nov 1977
7. L.H. Dixon, Average current-mode control of switching power supplies, in Unitrode Power
Supply Design Seminar Handbook, (1990), pp. 5.1–5.14
8. N. Mohan, Power electronics circuits: an overview, in Proceedings of IEEE Industrial
Electronics Society Conference, (1988), pp. 522–527
9. N. Mohan, W.P. Robbins, P. Imbertson, T.M. Undeland, R.C. Panaitescu, A.K. Jain, P. Jose,
T. Begalke, Restructuring of first courses in power electronics and electric drives that
integrates digital control. IEEE Trans. Power Electronics 18, 429–437 (2003)
10. R.D. Middlebrook, S. Cuk, A general unified approach to modeling switching-converter
power states, in Proceedings of IEEE Power Electronics Specialists Conference, (1976),
pp. 73–86
11. D.Y. Chen, H.A. Owen, T. G. Wilson, Computer Aided design and graphics applied to the
study of inductor-energy-storage dc-to-dc electronic power converters. IEEE Trans.
Aerosp. Electron. Syst. AES-9, 585 (1973)
12. P. Burger, Analysis of a class of pulse modulated dc-to-dc power converters. IEEE Trans. Ind.
Elect. Contr. Instrum. IECI-22, 104 (1975)
13. W.H. Lau, H. Chung, C.M. Wu, N.K. Poon, Realization of digital audio amplifier using
zero-voltage-switched PWM power converter. IEEE Trans. Circ. Syst. Part I. 47(3), 303–311
(2000)
14. C. Ni, T. Tetsuo, Adaptive constant on-time (D-CAP™) control study in notebook
applications, Texas Instruments, Application Report SLVA281B, July, 2007
15. R. Redl, J. Sun, Ripple-based control of switching regulators an overview. IEEE Trans. Power
Electron. 24, 2669–2680 (2009)
16. J. Li, Current-mode control: modeling and its digital application. Ph.D. thesis, Virginia
Polytechnic Institute and State University, 2009
17. J. Sun, Characterization and performance comparison of ripple-based control for voltage
regulator modules. IEEE Trans. Power Electron. 21, 346–353 (2006)
18. W. Huang, A new control for multi-Phase buck converter with fast transient response, in
Proceedings of IEEE Applied Power Electronics Conference and Exposition Conference,
(2001), pp. 273–279
19. J. Li, F.C. Lee, Modeling of V2 current-mode control, in Proceedings of IEEE Applied Power
Electronics Conference and Exposition Conferences, (2009), pp. 298–304
20. S.J. Wang, Y.H. Lee, Y.C. Lai, K.H. Chen, Quadratic differential and integration technique in
V2 control buck converter with small ESR capacitor, in Proceedings of IEEE Custom
Integrated Circuits Conference, (2009), pp. 211–214
21. R. Redl, G. Reizik, Switched noise filter for the buck converter using the output ripple as the
PWM ramp, in Proceedings of IEEE Appl. Power Electron. Conference, (2005), pp. 918–924
144 5 Adaptive On-time Control Circuit for Buck Converters

22. J. Li, F.C. Lee, New modeling approach and equivalent circuit representation for current-
mode control. IEEE Trans. Power Electron. 1218–1230 (2010)
23. Y.H. Lee, S.J. Wang, K.H. Chen, Quadratic differential and integration technique in V2
control buck converter with small ESR capacitor. Proc. IEEE Trans. Power Electron. 25, 829–
838 (2010)
24. M.Y. Yen, P. Mok, A constant frequency output-ripple voltage-based buck converter without
using large ESR capacitor. IEEE Trans. Circ. Syst. 55, 748–752 (2008)
25. K.Y. Cheng, F. Yu, P. Mattavelli, F.C. Lee, Characterization and performance comparison of
digital V2-type constant on-time control for buck converters, IEEE Cont. Model. Power
Electronics Conf. 1–6 (2010)
26. A.V. Petershevs, S.R. Sanders, Digital multimode buck converter control with loss-minimizing
synchronous rectifier adaptation, IEEE Trans. Power Electron. 21(6), 1588–1599 (2006)
27. S. Angkititrakul, H. Hu, Design and analysis of buck converter with pulse-skipping
modulation, in Proceedings of IEEE Power Electronics Specialists Conference, (2008),
pp. 1151–1156
28. X. Zhou, M. Donati, L. Amoroso, F.C. Lee, Improved light-load efficiency for synchronous
rectifier voltage regulator module. IEEE Trans. Power Electronics 15(5), 826–834 (2000)
29. C.L. Chen, W.L. Hsieh, W.J. Lai, K.H. Chen, C.S. Wang, A new PWM/PFM control
technique for improving efficiency over wide load range, in Proceedings of IEEE
International Conference on Electronics, Circuits and Systems, (Aug 2008), pp. 962–965
30. S. Kapat, S. Banerjee, A. Patra, Discontinuous map analysis of a DC-DC converter governed
by pulse skipping modulation, IEEE Trans. Circ. Syst. Part I. 57(7), 1793–1801 (2010)
31. W.W. Chen, J.F. Chen, T.J. Liang, L.C. Wei, J.R. Huang, W.Y. Ting, A novel quick response
of RBCOT with VIC ripple for Buck converter. IEEE Trans. Power Electron. 28, 4299–4308
(2013)
32. W.W. Chen, J.F. Chen, T.J. Liang, J.R. Huang, L.C. Wei, W.Y. Ting, Implementing dynamic
quick response with high-frequency feedback control of the deformable constant on-time
control for Buck converter on-chip. IET Power Electron. 6(4), 383–391 (2013)
33. W.W. Chen, J.F. Chen, T.J. Liang, S.F. Hsiao, J.R. Huang, W.Y. Ting, Improved transient
response using HFFC circuit of the CCRCOT with native AVP design for voltage regulators.
IET Power Electron. 6, 1948–1955 (2013)
34. W.W. Chen, J.F. Chen, T.J. Liang, J.R. Huang, W.Y. Ting, Improved transient response using
HFFC in current-mode CFCOT control for buck converter, in Proceedings IEEE
International Conference on Power Electronics and Drive Systems (PEDS), (2013),
pp. 546–549
Chapter 6
Ripple-Based Constant Frequency
On-time Control Circuit with Virtual
Inductor Current Ripple for Buck
Converters

6.1 Challenges for Adaptive On-time Control Circuits


for Buck Converters

Ceramic capacitors have many advantages for power converter applications.


Ceramic capacitors are non-polarized. Electrolytic capacitors are polarized, result-
ing in damage or even explosion when subjected to high reversed polarity pulses or
if they are mounted incorrectly. Ceramic capacitors have a wider frequency
bandwidth and lower impedance than tantalums, making them more effective in
suppressing noise in power lines. In the past, power converters needed two
capacitors: a high-value electrolytic capacitor and a low-value ceramic capacitor. At
present, high-value ceramic capacitors of the same size can be used. A single
ceramic capacitor, which offers a better frequency response than the combination of
electrolytic and ceramic capacitors, provides the required frequency response over a
wide frequency band. Today’s ceramic capacitors can do the entire task. In addition,
ceramic capacitors have other advantages, particularly at high frequencies.
Their ESR and ESL are much lower than those of tantalums. The ESR of ceramic
capacitors is substantially lower than that of tantalum equivalents. A low ESR
prevents overheating of the device and circuit, thus increasing the overall reliability.
Given these advantages, ceramic capacitors are highly suitable and preferred for
many applications, such as digital cameras, netbooks, smartphones, and tablet
computers due to their small size, low output voltage ripple, and high reliability
requirements. Moreover, ceramic capacitors do not need temperature derating. In a
circuit operating at 25 V and full-rated temperature, 25 V ceramic capacitors can be
selected with full confidence.
However, ceramic capacitors contain a small output voltage ripple with a low
ESR, which results in a small RCO*CO time constant that makes it difficult to meet
the critical design condition in Eq. (6.1). Thus, a small time constant results in an
instability problem of subharmonic oscillations. The first challenge in adaptive
on-time control circuits is the absence of virtual inductor current ripple to add to

© Springer Nature Singapore Pte Ltd. 2018 145


W.-W. Chen and J.-F. Chen, Control Techniques for Power
Converters with Integrated Circuit, Power Systems,
https://doi.org/10.1007/978-981-10-7004-4_6
146 6 Ripple-Based Constant Frequency On-time Control Circuit with …

feedback voltage VFB. This absence results in time-delay effects in the control
loop. Thus, these control circuits still suffer from subharmonic oscillations with
ceramic capacitors [1–13].

TON
RCO  CO [ ð6:1Þ
2

The input signals of the comparator are reference voltage VREF and feedback
voltage VFB. These signals can determine the output signal VTRIG of the comparator.
Reference voltage VREF is generally designed to be a constant voltage less than 1 V,
and it is without voltage ripple due to the regulation of the required and designed
output voltage. If reference voltage VREF has a voltage ripple, the output voltage
would become large and cannot meet the original design. A low voltage ripple of
feedback voltage VFB can result in poor noise immunity between reference voltage
VREF and feedback voltage VFB. Voltage signal VLX easily suffers from the jitter
phenomenon under poor noise immunity, which results in an unfixed switching
frequency. A low-accuracy switching frequency causes the power converter or
output device to encounter an error in operation. Hence, the second challenge in
adaptive on-time control circuits is the low voltage ripple of feedback voltage VFB,
which results in the jitter phenomenon under poor noise immunity between refer-
ence voltage VREF and feedback voltage VFB. A virtual inductor current ripple can
be added to feedback voltage VFB to achieve improved noise immunity in the
control loop. The virtual inductor current ripple exerts no impact on the output
voltage ripple because the virtual inductor current ripple is directly added to
feedback voltage VFB.
Voltage signal VLX is usually measured to determine its jitter performance, as
shown in Fig. 6.1. Figure 6.1 shows voltage signal VLX with the jitter phenomenon
under poor noise immunity. To determine the jitter performance, one pulse of
voltage signal VLX can be triggered on the rising and falling edges by enabling a

Fig. 6.1 The voltage signal V


VLX with a jitter phenomenon
at the worse noise immunity TMeasured
VLX VIN

0
t
TON TOFF_MIN TJitter TON

TOFF_MAX
6.1 Challenges for Adaptive On-time Control Circuits for … 147

signal accumulation function of the oscilloscope. Then, the next pulse of the trig-
gered pulse can contain a TMeasured time. If voltage signal VLX has the jitter phe-
nomenon, then TMeasured time has a longer width than TON time because TMeasured
time is equal to the sum of TJitter time and TON time by Eqs. (6.2) and (6.3). The
width of TMeasured time is equal to that of TON time when voltage signal VLX does
not exhibit the jitter phenomenon. With the jitter phenomenon, the maximum duty
cycle DMAX depends on TOFF_MIN, and the minimum duty cycle DMIN depends on
TOFF_MAX. DMAX and DMIN are calculated by Eqs. (6.4) and (6.5). Moreover, if the
duty cycle considers the jitter phenomenon, the duty cycle can be changed and
shown as Eq. (6.6). In general, the percentage of jitter (%Jitter) is used to estimate
the jitter performance. %Jitter depends on the D, DMAX, and DMIN by Eq. (6.7). The
specification of %Jitter is an uncommon definition in IC products’ datasheet, but the
internal specification of %Jitter provided by IC Design Company is less than 15%.

TJitter ¼ TMeasured  TON ð6:2Þ

TOFF MIN ¼ TOFF MAX  TJitter ð6:3Þ

TON
DMAX ¼ ð6:4Þ
TOFF MIN þ TON

TON
DMIN ¼ ð6:5Þ
TOFF MAX þ TON

TON
D¼ ð6:6Þ
TOFF MIN þ TON þ 0:5  ðTJitter Þ

DMAX  DMIN
%Jitter ¼ ð6:7Þ
D

Figure 6.2 shows the experimental results of voltage signal VLX and voltage
signal VOUT with the jitter phenomenon under poor noise immunity for adaptive
on-time control circuits. These experimental results were measured to trigger the
rising edge of the first pulse by enabling a signal accumulation function of an
oscilloscope. The next pulse of the triggered pulse can contain TMeasured time with
the jitter phenomenon. TMeasured time is 664 ns, which includes many times of TON
time. Pure TON time is the first pulse, and it is 200 ns. TJitter time is 464 ns. To
determine the percentage of jitter (%Jitter), TOFF_MIN time can be measured based on
the experimental results. TOFF_MIN time is 2216 ns, so TOFF_MAX time can be
calculated as 2680 ns. Based on Eqs. (6.3)–(6.5), D, DMAX, and DMIN can be
calculated as 7.553, 8.278, and 6.944%, respectively. Then, %Jitter can be obtained
as 17.658%. This value is beyond the internal specification because it is larger than
15%.
148 6 Ripple-Based Constant Frequency On-time Control Circuit with …

VOUT

TMeasured
664ns

VLX TOFF_MIN=2216ns

TON=200ns

Fig. 6.2 Experimental results of the voltage signal VLX and the voltage signal VOUT with a jitter
phenomenon at the worse noise immunity for the adaptive on-time control circuits

6.2 Implemented Control Circuits of the Victual Inductor


Current Ripple

The victual inductor current ripple can be used to solve and improve the challenges
in adaptive on-time control circuits, such as a small time constant that results in the
instability problem with subharmonic oscillations and a low voltage ripple of
feedback voltage VFB that results in the jitter phenomenon under poor noise
immunity between reference voltage VREF and feedback voltage VFB. Hence, the
victual inductor current ripple must be added to feedback voltage VFB.
Adding the inductor current ripple to feedback voltage VFB is a good means to
overcome loop stability problems because the inductor current ripple depends on
the duty cycle, so it is a synchronous signal with feedback voltage VFB ripple. In
addition, the inductor current ripple is a sawtooth wave to be added to feedback
voltage VFB to increase the noise immunity and avoid time-delay effects in the
control loop. For a large ESR such as that in electrolytic capacitors, feedback
voltage VFB ripple is similar to the inductor current ripple. For this reason, the
control circuit also requires duty cycle information to generate the victual inductor
current ripple.
In general, the control circuit uses two types of control signals to generate a
victual inductor current ripple. One type of control signal is voltage signal VLX, and
6.2 Implemented Control Circuits of the Victual Inductor Current Ripple 149

VLX-VOUT VIN-VOUT

0
-VOUT
t

VLX VIN

0
t

Fig. 6.3 Two types control signals to generate a victual inductor current ripple of the voltage
signal VLX and the voltage signal VLX-VOUT

the other type is the voltage drop between voltage signal VLX and output voltage
VOUT (Fig. 6.3) because signal VLX or the voltage drop between voltage signal VLX
and output voltage VOUT have duty cycle information. These two types of control
signals have the same peak-to-peak voltage, so they can be used to generate a
victual inductor current ripple.
Moreover, the pulse control signal can use a low-pass filter to generate a saw-
tooth wave similar to the inductor current ripple. This sawtooth wave is called a
victual inductor current ripple. The low-pass filter is placed between voltage signal
VLX and output voltage VOUT or in voltage signal VLX only. Two types of control
circuits are typically used with the virtual inductor current ripple of adaptive
on-time control circuits using a low-pass filter to connect voltage signal VLX and
output voltage VOUT. One type is inductor’s DCR current sensing control circuit
[14–19]. Another one type of control circuit with virtual inductor current ripple of
adaptive on-time control circuits using a low-pass filter to connect voltage signal
VLX is typically used; this type is ripple-based control circuit [1–5, 20–26].
The DCR current sensing control circuit is an alternative to a sense resistor, so a
sense resistor does not need to be added. The DCR current sensing control circuit
utilizes the parasitic resistance of an inductor to measure the current to the load. It
remotely measures the current through an energy-storing inductor of a switching
regulator circuit. Figure 6.4 shows the DCR current sensing control circuit of the
adaptive on-time control circuits for buck converter. SA and SB are the switches, L is
the output inductor, and RCO is the ESR of the output capacitor CO. The current
source IOUT is the output load current, and RD1 and RD2 are the feedback resistors to
determine the output voltage. VFB is a feedback voltage. The reference voltage VREF
150 6 Ripple-Based Constant Frequency On-time Control Circuit with …

VIN SA VLX L DCR VOUT


VUG
IL IOUT
VLG RCO
SB RX VLPFCX
CO
RD1
Driver IX

VFB
On-Time VTRIG
Generator VSUM
VDD VDD
VOUT G1 G2 VREF RD2
VIN VOUT VLPF
RSUM

Fig. 6.4 The DCR current sensing control circuit of the adaptive on-time control circuits for buck
converter

is created inside the IC. The on-time generator circuit samples the VIN signal and
VOUT signal to adjust the on-time width to control the driver circuit and achieve the
voltage regulation. A low-pass filter consists of a series of resistor RX and capacitor
CX that are connected between voltage signal VLX and output voltage VOUT.
The most accurate sensing of the voltage across DCR is achieved by matching
the time constant of resistor RX and capacitor CX filter with inductor L and its DCR
time constant. The time constant of inductor L and its DCR are much larger than
switching period TS. The inductor current ripple contains a sawtooth wave.
Switching period TS is usually in the order of microseconds, given a switching
frequency in the order of a few hundred kHz to a few MHz. The time constant of
inductor L and its DCR are typically in the order of milliseconds, so the inductor
current ripple always has a sawtooth wave. If the time constant of resistor RX and
capacitor CX filter are selected to be equal to the time constant of inductor L and its
DCR by Eq. (6.8), then voltage signal VCX of capacitor CX is directly proportional
to the inductor current ripple, and voltage signal ripple VCX is a virtual inductor
current ripple. If the time constant of resistor RX and capacitor CX filter is not equal
to the time constant of inductor L and its DCR, then the virtual inductor current
ripple cannot obtain the same slew rate as the inductor current ripple.

L
RX  CX ¼ ð6:8Þ
DCR

Voltage signal VCX can be calculated by Eqs. (6.9)–(6.11). If a user wants to


know DC voltage signal VCX, it can be calculated by Eq. (6.12). Voltage signal VCX
is also equal to the drop between voltage signal VLPF and output voltage VOUT.
6.2 Implemented Control Circuits of the Victual Inductor Current Ripple 151

A real implemented IC uses two voltage-controlled current sources to add to the


virtual inductor current ripple in feedback voltage VFB, or the virtual inductor
current ripple is subtracted in reference voltage VREF, as shown in Fig. 6.4, because
the same intent can be achieved. G1 and G2 are the gain of the voltage-controlled
current source. They can directly control the amplitude of sum voltage ripple VSUM.

IL  ðs  L þ DCRÞ ¼ IX  RX þ VCX ð6:9Þ

IX ¼ s  CX  VCX ð6:10Þ

IL  ðs  L þ DCRÞ ¼ VCX  ð1 þ s  RX  CX Þ ð6:11Þ

VCX ¼ IL  DCR ð6:12Þ

Figure 6.5 shows these control signals with the DCR current sensing control
circuit of adaptive on-time control circuits for buck converter. Reference voltage
VREF does not have a ripple-based circuit, so VREF maintains a constant DC value.

VLX-VOUT VIN-VOUT

0
-VOUT
t
VLPF-VOUT IL_Max DCR
0
IL_Min DCR
t
VFB
VSUM VREF-IL_Min DCR
VREF
VREF-IL_Max DCR
t
VTRIG

Fig. 6.5 Control signals with the DCR current sensing control circuit of the adaptive on-time
control circuits for buck converter
152 6 Ripple-Based Constant Frequency On-time Control Circuit with …

The drop between voltage signal VLPF and output voltage VOUT can be controlled to
become a sawtooth wave by the DCR current sensing control circuit. The DC level
of VA is equal to zero voltage, and VA can be subtracted in reference voltage VREF to
increase the noise immunity and loop stability. If voltage VSUM is larger than
feedback voltage VFB and control signal VTRIG is changed from a low to high level,
then control signal VTRIG is needed to determine the final on-time width of TON.
Output voltage VOUT maintains the same voltage ripple even when the adaptive
on-time control circuits use the DCR current sensing control circuit. The DCR
current sensing control circuit is used not only to increase voltage ripple VREF but
also to maintain the same efficiency and output voltage ripple of the power
converter.
However, the time constant of resistor RX and capacitor CX filter matching and
DCR temperature dependence are important constraints that must be addressed to
achieve a high level of current-sensing accuracy across variations in loads and
operating temperatures. The DCR current-sensing control circuit can exhibit high
performance, but the DCR parameter drifts when the temperature changes. Thus,
DCR requires temperature compensation for current sensing in point-of-load reg-
ulator applications. In most IC products, the negative temperature coefficient
(NTC) resistor is a good means to achieve temperature compensation. The NTC
resistor should be added to the resistor network to monitor the inductor current for
each phase and to achieve DCR thermal compensation [27–29]. Meanwhile, the
DCR current-sensing control circuit needs to add two pins to achieve this function
because the real implemented IC needs to sample voltage signal VLPF and output
voltage VOUT individually. To achieve high accuracy in inductor current sensing, an
auto tuning function must be implemented to overcome the impact of the inductor
and its DCR tolerances. Thus, high-accuracy inductor current sensing with the DCR
current-sensing control circuit is widely applied, such as in CPU voltage regulator
applications.
The ripple-based control circuit is a type of control circuit with virtual inductor
current ripple of adaptive on-time control circuits using a low-pass filter consisting
of a series of resistor RLPF and capacitor CLPF that are connected to voltage signal
VLX. The ripple-based control circuit is also an alternative to a sense resistor, so
adding a sense resistor is unnecessary. The ripple-based control circuit consists of
RLPF, CLPF, and DC value extractor.
The ripple-based control circuit of the adaptive on-time control circuit with
virtual inductor current ripple for buck converter is as shown in Fig. 6.6. SA and SB
are the switches, L is the output inductor, and RCO is the ESR of the output
capacitor CO. The current source IOUT is the output load current, and RD1 and RD2
are the feedback resistors to determine the output voltage. VFB is a feedback
voltage. The reference voltage VREF is created inside the IC. The on-time generator
circuit samples the VIN signal and VOUT signal to adjust the on-time width to control
the driver circuit and achieve the voltage regulation.
The purpose of a virtual inductor current ripple is to alleviate the instability
problem because it enhances the effect of ESR voltage ripple in the feedback
voltage. This circuit provides improved system stability, especially in all ceramic
6.2 Implemented Control Circuits of the Victual Inductor Current Ripple 153

SA VLX L
VIN VOUT
VUG
IL IOUT
VLG SB Rco
Co
RD1
Driver

VSUM VFB
VTRIG K1
On-Time
Generator
VREF RD2

VIN VOUT K2

VVIC (0)
Ripple-Based Control Circuit
VLX RLPF VCLPF

CLPF
DC value
extractor VCANCEL

Fig. 6.6 The ripple-based control circuit of the adaptive on-time control circuit with virtual
inductor current ripple for buck converter

capacitors for output capacitors, which normally have relatively low ESR values. In
a ripple-based control circuit, voltage VCLPF can be obtained by integrating voltage
signal VLX through RLPF and CLPF integrators. Voltage VCLPF is a sawtooth wave,
and its DC level is equal to the output voltage. Hence, VCLPF cannot be directly
added to feedback voltage VFB. The DC value of VCLPF is removed by the DC value
extractor to generate the same voltage ripple with a zero DC value. One type of
implemented circuit of the DC value extractor can be used with another low-pass
filter to generate voltage VCANCEL, which is similar to output voltage VOUT. Voltage
VCLPF can subtract voltage VCANCEL to obtain voltage VVIC with the same ripple
voltage as VCLPF. However, its DC value is equal to zero voltage. Voltage VVIC is
added to feedback voltage VFB to enhance the ESR voltage ripple.
The low-pass filter for RLPF and CLPF should be designed close to the resonant
frequency of output inductor L and output capacitor CO by Eq. (6.13). The equation
needs to consider the minimum on-time TON by Eq. (6.14). Hence, the designs of
resistor RLPF and capacitor CLPF need to meet the criteria for minimum on-time TON
with a ripple-based control circuit. The design consideration of resistor RLPF and
capacitor CLPF with a ripple-based control circuit is the similar to that of the D-CAP
control circuit.
154 6 Ripple-Based Constant Frequency On-time Control Circuit with …

pffiffiffiffiffiffiffiffiffiffiffiffi
RLPF  CLPF ¼ L  CO ð6:13Þ

L  CO TON
[ ð6:14Þ
RLPF  CLPF 2

The main design consideration of the DC value extractor is to generate voltage


VCANCEL. This voltage obtains the same DC value as VOUT. The time constant of
the low-pass filter for the DC value extractor can affect its voltage ripple. If the time
constant is too small, voltage VCANCEL exhibits a large ripple. As a result, voltage
VVIC ripple cannot avoid the time-delay effects in the control loop. For this reason
the time constant of the low-pass filter for the DC value extractor is approximately
the time constant of resistor RLPF and capacitor CLPF filter to avoid the time-delay
effects in the control loop. However, if the user considers increasing the transient
response, the time constant of the low-pass filter for the DC value extractor can be
designed to equal a half or less the time constant of resistor RLPF and capacitor CLPF
filter because the virtual inductor current ripple is traded off to avoid time-delay
effects in the control loop and to increase the transient response.
Figure 6.7 shows these control signals with the ripple-based control circuit of
adaptive on-time control circuits for buck converters. The original feedback voltage
VFB_ORIGINAL does not have a ripple-based control circuit, so VFB_ORIGINAL has a
low voltage ripple, resulting in time-delay effects in the loop with a low ESR. Sum
voltage VSUM is the sum voltage of voltage VVIC and original voltage VFB_ORIGINAL
with a ripple-based control circuit. VSUM has a larger voltage ripple than the original
voltage VFB_ORIGINAL to increase noise immunity and loop stability because VSUM
ripple is similar to voltage VCLPF ripple.
Output voltage VOUT maintains the same waveform even when adaptive on-time
control circuits employ a ripple-based control circuit. The ripple-based control
circuit is used not only to increase the VSUM ripple but also to maintain the same
efficiency and output voltage ripple in the power converter.
A real implemented IC directly uses sum voltage VSUM as the input terminal of
the comparator, and the output voltage can be regulated and calculated by
Eq. (6.15). Output voltage VOUT is smaller than the required voltage, so the issue of
the DC level of output voltage VOUT being unequal to the required voltage needs to
be solved.
 
RD1
VOUT ¼ ðVFB  0:5  DVRIPPLE Þ  1 þ ð6:15Þ
RD2

If voltage VREF is larger than sum voltage VSUM, then control signal VTRIG is
changed from a low to high level. Control signal VTRIG is needed to determine the
final on-time width of TON. Moreover, a ripple-based control circuit can be
embedded by a real implemented IC. Voltage signal VLX also exists in IC, so no
extra IC pin is needed to implement the modified control [1–13] because the
ripple-based control circuit does not need to sample the output voltage signal. In
6.2 Implemented Control Circuits of the Victual Inductor Current Ripple 155

VLX VIN

0
t
VFB_ORIGINAL
VFB
t
VCLPF
VRIPPLE
VOUT

t
VCANCEL
VOUT
t
VVIC
VRIPPLE
0

t
VSUM
VFB+0.5 VRIPPLE
VFB
VFB-0.5 VRIPPLE
t
VFB+0.5 VRIPPLE
VSUM
VFB
VREF
VREF
(=VFB-0.5 VRIPPLE)
t
VTRIG

0
t

Fig. 6.7 Control signals with the ripple-based control circuit of the adaptive on-time control
circuits for buck converter
156 6 Ripple-Based Constant Frequency On-time Control Circuit with …

addition, the real implemented IC should have a trimmed code option to change the
suitable parameters of resistor RLPF, capacitor CLPF, and DC value extractor to meet
different inductors and operation conditions.

6.3 On-time Generator Circuit of the Constant Frequency


On-time Control Circuits for Buck Converters

On-time generator circuit of the constant frequency on-time control circuit is as


shown in Fig. 6.8. The TON pin is in one of the IC, and the resistor R1 can be
placed between TON pin and input voltage to determine the width of on-time. VOUT
needs to be decided the peak voltage of capacitor C1 by Eq. (6.16). The output
voltage VOUT divided by the input voltage VIN is a duty cycle, and R1, C1, and G1
are constant values, as shown in Eq. (6.17). G1 is a gain of current-controlled
current source. Even if the input voltage VIN and output voltage VOUT are changed,
the system still maintains the same switching frequency [1–5].

VOUT R1  C1
TON ¼  ð6:16Þ
VIN G1

VOUT
TON ¼  TS ð6:17Þ
VIN

Constant frequency on-time control circuit is a different from the conventional


COT control circuit because it avoids the generation of the same on-time width at a
high VOUT. Regardless of changes in VIN or VOUT, the conventional COT control
circuit still generates the same on-time width. If the conventional COT control
circuit wants to regulate a high VOUT and needs to generate more on-time pulses, an
increase in switching loss occurs. Thus, the constant frequency on-time control
circuit is suitable for a wide range output voltage.

VDD

VIN R1 TON
G1
On-Time

VTRIG
S1 C1 VOUT

Fig. 6.8 On-time generator circuit of the constant frequency on-time control circuit
6.3 On-time Generator Circuit of the Constant Frequency On-time … 157

Switching Frequency vs. Input Voltage


Output Voltage=3.3V, Output Current=0.1A
520
Switching Frequency (kHz) 480
(1)
440
400
360
320
280 (2)
240
200 (2) R1=510 kΩ
(1) R1=320 kΩ
160
2 4 6 8 10 12 14 16 18 20 22
Input Voltage (V)

Fig. 6.9 Experimental results of the switching frequency versus the input voltage for constant
frequency on-time control circuit

Experimental results of the switching frequency versus the input voltage for
constant frequency on-time control circuit are as shown in Fig. 6.9. The operation
conditions are set at 3.3 V output voltage and 0.1 A output current to measure the
switching frequency for 4–21 V input voltage. When R1 is equal to 510 kX, con-
necting VIN and TON pin, the switching frequency is 255 kHz, regardless of
changes in VIN. If R1 is reduced to 320 kX, the switching frequency changes from
255 to 444 kHz and the resistor R1 directly affects the on-time width. Between the
resistor R1 and the switching frequency are the inverse relationships.
Experimental results of the inductor current ripple versus the input voltage for
constant frequency on-time control circuit are as shown in Fig. 6.10. The operating
conditions are set 3.3 V output voltage and 0.1 A output current to measure the
inductor current ripple for 4–21 V input voltage. If the inductance is invariant, then
the inductor current ripple is related to the voltage drop between the input voltage
and output voltage and is also dependent on the on-time TON, as shown by
Eq. (6.18). By substituting Eq. (6.17) into Eq. (6.18), the inductor current ripple is
obtained as Eq. (6.19). The inductor current ripple increases as the input voltage is
increased. Hence, it does not maintain a constant value. When R1 is increased from
320 to 510 kX, the on-time width is also increased, resulting in a larger inductor
current ripple.

VIN VOUT
DIL ¼  TON ð6:18Þ
L
VOUT ð1 - ðVOUT =VIN ÞÞ
DIL ¼  TS ð6:19Þ
L
158 6 Ripple-Based Constant Frequency On-time Control Circuit with …

Inductor Current Ripple vs. Input Voltage


Output Voltage=3.3V, Output Current=0.1A
4.5
Inductor Current Ripple (A) 4
(2)

3.5

2.5 (1)
2

1.5 (2) R1=510 kΩ


(1) R1=320 kΩ
1
2 4 6 8 10 12 14 16 18 20 22
Input Voltage (V)

Fig. 6.10 Experimental results of the inductor current ripple versus the input voltage for constant
frequency on-time control circuit

6.4 Comparison of Quick Dynamic Response


and Conventional Quick Response of the On-time
Generator Circuit

Two types of improved transient response circuits are typically used, namely,
internal fixed by IC design and external setting by component or voltage source.
The internal fixed by IC design usually uses the feedback voltage VFB to trigger the
quick response function, which can provide a long on-time width to control the
driver circuit. However, the generator circuit of quick response has a very similar
on-time generator circuit. Figure 6.11 shows the internal fixed quick response of the
on-time generator circuit for constant frequency on-time control circuit [1–5].
The internal fixed quick response needs to add a voltage source VREF2. The
VREF2 is usually designed at 80–90% of VREF to prevent the occurrence of under
voltage protection. If the feedback voltage VFB is lower than VREF2, the output
signal of the comparator changes from high-level to low-level, which can turn off
the switch of S2. The voltage source VREF1 is set in series connection with VOUT to
increase the on-time width. VREF1 is designed by the maximum duty cycle limited
[1–5].
The main advantage of this implementation is that no extra pin in the IC is
needed to achieve the quick response function. However, the design of the internal
fixed quick response can generate just one form of a long on-time width. Thus, this
circuit cannot dynamically change the on-time width during different load condi-
tions. Besides, the hysteresis of the comparator is difficult to design, especially for a
very fast load transient. If the hysteresis of the comparator is too small, it can induce
the system to erroneously trigger a long on-time width, thereby causing high output
6.4 Comparison of Quick Dynamic Response and Conventional Quick … 159

R1 TON G1
VIN VDD
On-Time

VTRIG S1 C1

VOUT

S2
VFB VREF1

VREF2

Fig. 6.11 Internal fixed quick response of the on-time generator circuit for constant frequency
on-time control circuit

voltage. If the hysteresis of the comparator is too large, it can also render the quick
response function unavailable.
External setting quick response of the on-time generator circuit for constant
frequency on-time control circuit is as shown in Fig. 6.12. The operation principle
of the external setting quick response requires the addition of a voltage source
VQRSET to determine the QR-time width. VQRSET is connected to the QRSET pin. If
the QR-time needs a long QR-time width, VQRSET should be larger than VOUT.
0 0
VOUT determines the peak voltage of C1. Thus, if VOUT is high, the peak voltage
0
increases, causing a long on-time width. VOUT is implemented to provide a
ripple-less voltage because it can keep the switching frequency constant without
0
under output voltage ripple and noise interference. VOUT is implemented to sample
the Phase signal through a second-order low-pass filter, producing a similar VOUT
0
signal. Therefore, VOUT is ripple-less and has slower transient response than the
VOUT signal.
The QR trigger circuit samples VOUT and uses the low-pass filter to cause VOUT
delay. This trigger circuit also needs the voltage signal VQRTH. In the steady-state
operation, the VOUT drop cannot trigger the QR-time. When this abrupt voltage drop
is lower than VQRTH, the QR trigger circuit changes from a high-level signal to a
low-level signal to turn OFF the S2 switch and then the on-time generator circuit
generates the QR-time width. The low-pass filter frequency for RLOW and CLOW
should be designed much smaller than the switching frequency which can avoid this
system to fail in its operation with QR-time. VQRTH can be designed in the IC or set
by the user through the QRTH pin. The main advantage of the external setting
160 6 Ripple-Based Constant Frequency On-time Control Circuit with …

VIN R1 TON G1
VDD
On-Time

VTRIG S1 C1 V'OUT

G2
VDD
QR-Time

VQRTH
VOUT
C2
S2
QRSET
RLOW

QR trigger VQRSET
CLOW
circuit

Fig. 6.12 External setting quick response of the on-time generator circuit for constant frequency
on-time control circuit

quick response is that it can depend on the load conditions to design the VQRSET
value, generating a suitable QR-time width in this system. However, this method
requires two extra IC pins to achieve the quick response function [1–5].
Quick dynamic response of the ripple-based constant frequency on-time control
circuit with virtual inductor current ripple for buck converter is proposed, as well as
quick dynamic response of on-time generator circuit for constant frequency on-time
control circuit, as shown in Fig. 6.13. An external setting device is used to design
these components, instead of a set-up voltage source. This quick dynamic response
does not require an extra pin to achieve the quick response function of IC. The
system consists of a series of the resistor Rq and capacitor Cq that are connected
between the output voltage VOUT and TON pin. It operates under the principle that
the high-pass filter filters high frequency signal to pass from VOUT to the TON pin
by Eq. (6.20). However, the frequency of the high-pass filter must be larger than or
equal the switching frequency so the steady state operation of the system is not
affected. On the other hand, the user may base the design of the resistor Rq and
capacitor Cq on the worst-case operation (maximum loading step) [1–5].
0 0
VOUT determines the peak voltage of C1. Thus, if VOUT is high, the peak voltage
0
increases, causing a long on-time width. VOUT is implemented to provide a
ripple-less voltage because it can keep the switching frequency constant without
0
under output voltage ripple and noise interference. VOUT is implemented to sample
6.4 Comparison of Quick Dynamic Response and Conventional Quick … 161

VIN

R1 I1

TON G1
VDD
On-Time

Rq

I2 VTRIG C1
S1 V'OUT
Cq

VOUT

Fig. 6.13 Quick dynamic response of the on-time generator circuit for constant frequency on-time
control circuit

the Phase signal through a second-order low-pass filter producing a similar VOUT
0
signal. Therefore, VOUT is ripple-less and has a slower transient response than VOUT.
1
FRC ¼  FS ð6:20Þ
2p  Rq  Cq

When light load quickly transforms to heavy load, VOUT momentarily drops
through the coupling by the capacitor Cq, which induces the resistor Rq to cause a
voltage drop. The voltage drop of Rq will form a current of I2. Thus, the resistor Rq
needs to be designed first because the current of I2 can directly change the width of
on-time. However, a longer width of on-time induces the converter to deliver more
energy from the input terminal to the output loading. If the resistor Rq has reduced,
the width of on-time becomes longer. The capacitor of Cq should be designed with
the frequency of the high-pass filter. The calculation of the Laplace transform
formula is shown in Eq. (6.21).

C1 0
TON ðsÞ ¼      VOUT ð6:21Þ
G1 ðVIN /R1 Þ þ sCq VOUT / 1 þ sCq Rq

The advantages of this quick dynamic response are as follows: (1) It clearly
generates a longer width of on-time that is proportional to the output voltage drop;
(2) Adaption to the width of on-time depends on the load conditions of the system;
(3) It only uses one on-time generator circuit; thus, it is very convenient to design
and apply; (4) It does not require an extra pin to achieve the quick response circuit
function of IC.
162 6 Ripple-Based Constant Frequency On-time Control Circuit with …

6.5 Experimental Results

The experimental results are shown to prove the feasibility and performance with
quick dynamic response of the ripple-based constant frequency on-time control
circuit with virtual inductor current ripple for buck converter. The specifications are
as follows:
(1) Input voltage (VIN): 12 V
(2) Output voltage (VOUT): 3.3 V
(3) Maximum output load current (IOUT): 18 A @ 1.2857 A/ls
(4) Switching frequency (FS): 255 kHz
(5) MOSFET (SA, SB): BSC0909NS * 3
(6) Feedback resistors (RD1, RD2): 68 and 12 kX
(7) Main inductor (L): IHLP4040DZER1R0MA1 (1 lH)
(8) Output capacitors (CO): 22lF/6.3 V (RCO: 3 mX)*13
(9) Reference voltage (VREF): 0.5 V
(10) Quick dynamic response: G1 = 1(A/A); R1 = 510 kX; C1 = 7.06 pF;
Rq = 510 X; Cq = 390 pF;
Figure 6.14 shows the chip layout of the control IC, in which the low-pass filter
of the virtual inductor current ripple occupies the area marked by the LP Filter with
a die size of 780 lm  780 lm.
Experimental results at the load transient with quick dynamic response (see
Fig. 6.13) and without quick response (see Fig. 6.8) are as shown in Fig. 6.15. The
blue-colored waveform represents without quick response for the output voltage

Fig. 6.14 Chip layout of the


control IC
DRIVER & Driver Logic

Bias
Comparator & LP Filter
Reference

780 μm
Protection

780 μm
6.5 Experimental Results 163

Fig. 6.15 Comparison of the


experimental results at the 396mV
load transient with quick
(1)
dynamic response and
without quick response
VOUT(200mV/20μs) 481mV
(2)

VOUT(200mV/20μs)

IOUT(10A/20μs)
(3)
(2) Without Quick Response
(1) Quick Dynamic Response

signal, whereas the red-colored waveform represents with quick dynamic response
for the output voltage signal. VOUT peak to peak value at the load transient with
quick dynamic response is 396 mV, and VOUT peak to peak value at the load
transient without quick response circuit is 481 mV. Thus, VOUT peak to peak value
with quick dynamic response is lower by 85 mV than without quick response,
which is useful to prevent the VOUT from dropping significantly.
The output voltage signals are measured based on the same output load current,
such as the black-colored waveform. The output load current is used as a function
generator signal to control the switch of MOSFET to implement the load transient,
which is faster than the electronics load as shown in Fig. 6.16.
Experimental results at the droop with quick dynamic response and without
quick response are as shown in Fig. 6.17. The blue-colored waveforms represent
without quick response circuit for the output voltage and VUG signal, whereas the
red-colored waveforms represent with quick dynamic response for the output

Fig. 6.16 Comparison of the Control Signal (5V) (20μs/div)


slew rate at the load transient
for the electronics load and
MOSFET implemented with a (1)
function generator signal Output Current (10A)

(2)
Output Current (10A)

(3)
(1) Control Signal from Function Generator(Black)
(2) MOSFET Implemented (Red)
(3) Electronics Load (Blue)
164 6 Ripple-Based Constant Frequency On-time Control Circuit with …

Fig. 6.17 Comparison of the


experimental results at the VUG(20V/4μs)
droop with quick response
dynamic and without quick (1)
response VUG(20V/4μs)

(2)

243mV VOUT(200mV/4μs)
(3)

VOUT(200mV/4μs)
335mV (4)

(2),(4)Without Quick Response


(1),(3) Quick Dynamic Response

voltage and VUG signal, whose signals are measured based on the same output load
current. VOUT peak to peak value at the droop with quick dynamic response is
243 mV, and VOUT peak to peak value at the droop without quick response circuit is
335 mV. Thus, VOUT peak to peak value with quick dynamic response is lower by
92 mV than without quick response, which is useful to prevent the VOUT from
dropping significantly. The setting time with quick dynamic response at the droop is
faster than without quick response, because VOUT at the droop with quick dynamic
response does not suffer an overshoot. On the other hand, the VUG signal with quick
dynamic response is obviously longer than without quick response circuit at the
droop, which enables the input power to deliver more energy to the output terminal.
Experimental results at the load release with quick dynamic response and
without quick response are as shown in Fig. 6.18. The blue-colored waveforms
represent without quick response for the output voltage and VUG signal, whereas the
red-colored waveforms represent with quick dynamic response for the output
voltage and VUG signal, whose signals are measured based on the same output load
current. VOUT at the load release with quick dynamic response varies closely
without quick response. The settling time at the load release with quick dynamic
response is also shorter than without quick response because VOUT does not suffer a
voltage drop.
To further understand the advantage and superiority of this quick dynamic
response, the plans based the same VOUT peak to peak value at the droop with quick
dynamic response on to increase additional 22 µF of ceramic output capacitors
without quick response.
The conditions are as follows:
Condition 1: Quick dynamic response
(1) Output capacitors CO: 22 lF/25 V (RCO: 3 mX)  13
(2) Quick dynamic response: G1 = 1(A/A); R1 = 510 kX; C1 = 7.06 pF; Rq = 510
X; Cq = 390 pF;
6.5 Experimental Results 165

Fig. 6.18 Comparison of the VUG(20V/10μs)


experimental results at the
load release with quick (1)
dynamic response and
without quick response VUG(20V/10μs)

(2)

VOUT(200mV/10μs) (3)

VOUT(200mV/10μs) (4)

(2),(4)Without Quick Response


(1),(3) Quick Dynamic Response

Condition 2: Without quick response using extra 17  22 µF ceramic capacitors


(1) Output capacitors CO: 22 lF/25 V (RCO: 3 mX)  30
Based on the same VOUT peak to peak value at the droop with quick dynamic
response, it needs to add how many 22 µF of output ceramic capacitors without
quick response. Figure 6.19 shows a comparison of the experimental results at the
load transient with quick dynamic response and without quick response using extra
17  22 µF of ceramic output capacitors. The blue-colored waveform represents
without quick response with 30  22 µF ceramic output capacitors for the output
voltage signal, whereas the red-colored waveform represents with quick dynamic
response with 13  22 µF ceramic output capacitors for the output voltage signal,
whose signals are measured based on the same output load current. Thus, the large
output capacitance can reduce the VOUT ripple at the load transient, but the cost and
size of the output capacitance need to be sacrificed. On the other hand, the large

Fig. 6.19 Comparison of the


experimental results at the
load transient with quick 243mV
dynamic response and (1)
without quick response using
VOUT(200mV/20μs)
extra 17  22 µF of ceramic
output capacitors 243mV (2)

VOUT(200mV/20μs)

IOUT(10A/20μs)

(3)
(2) Without Quick Response 22μF*30
(1) Quick Dynamic Response 22μF*13
166 6 Ripple-Based Constant Frequency On-time Control Circuit with …

output capacitance induces the system loop response slowly. The setting time at
load transient with quick dynamic response is faster than without quick response
because VOUT at the droop with quick dynamic response does not suffer an over-
shoot and VOUT at the load release with quick dynamic response does not suffer a
voltage drop.
Based on the same VOUT peak to peak value at the droop with quick dynamic
response, it needs to set additional 17  22 µF of ceramic output capacitors
without quick response. Thus, quick dynamic response at this operating condition,
which can save 17  22 µF ceramic output capacitors, regardless of the cost and
size of the output capacitance, both of which are the very benefits of the circuit
design.
Experimental results at the droop with quick dynamic response and without
quick response using extra 17  22 µF of ceramic output capacitors are as shown
in Fig. 6.20. The blue-colored waveforms represent without quick response with
30  22 µF ceramic output capacitors for the output voltage and VUG signal,
whereas the red-colored waveforms represent with quick dynamic response with
13  22 µF ceramic output capacitors for the output voltage and VUG signal, whose
signals are measured based on the same output load current. The setting time with
quick dynamic response at the droop is 38 µs, which is faster by 25.5 µs than
without quick dynamic response, because VOUT at the droop with quick dynamic
response does not suffer an overshoot. Thus, the large output capacitance can hold
the VOUT drop and induce the VOUT ripple small. However, it may cause the system
to obtain a slow transient response.
Figure 6.21 shows a comparison of the experimental results at the load release
for the quick dynamic response and without quick response. The blue-colored
waveforms represent without quick response with 30  22 µF ceramic output
capacitors for the output voltage and the VUG signal, whereas the red-colored
waveforms represent the quick dynamic response with 13  22 µF ceramic output
capacitors for the output voltage and the VUG signal, whose signals are measured
based on the same output load current.

Fig. 6.20 Comparison of the


experimental results at the VUG(20V/10μs)
droop with quick dynamic
response and without quick (1)
response using extra 17  22
VUG(20V/10μs)
µf of ceramic output
capacitors (2)

VOUT(200mV/10μs)
38µs (3)

243mV VOUT(200mV/10μs)
63.5µs (4)

243mV (2),(4)Without Quick Response 22µF*30


(1),(3) Quick Dynamic Response 22µF*13
6.5 Experimental Results 167

Fig. 6.21 Comparison of the


experimental results at the VUG(20V/10μs)
load release with quick
dynamic response and (1)
without quick response using VUG(20V/10μs)
extra 17  22 µF of ceramic
output capacitors (2)

VOUT(200mV/10μs)
(3)

VOUT(200mV/10μs) (4)
(2),(4)Without Quick Response 22µF*30
(1),(3) Quick Dynamic Response 22µF*13

Thus, the large output capacitors can reduce the peak voltage of VOUT at the load
release, but the cost and size of the output capacitors need to be sacrificed. On the
other hand, the large output capacitors induce the system loop response slowly. The
setting time at load release with the quick dynamic response is faster than without
quick response.

6.6 Summary

This chapter proposes quick dynamic response of the ripple-based constant fre-
quency on-time control circuit with virtual inductor current ripple for buck con-
verter. Using the capacitor and resistor in series to filter VOUT at load transient to
dynamically change the width of on-time, the input power delivers more energy to
the output terminal to prevent VOUT from dropping significantly.
Experimental results confirm quick dynamic response which can reduce in cost
and size of the ceramic output capacitance and improve in transient response.
Moreover, the proposed quick dynamic response has a very simple structure and
design of the ripple-based constant frequency on-time control circuit with virtual
inductor current ripple for buck converter.

References

1. W.W. Chen, J.F. Chen, T.J. Liang, L.C. Wei, J.R. Huang, W.Y. Ting, A novel quick response
of RBCOT with VIC ripple for buck converter. IEEE Trans. Power Electron. 28, 4299–4308
(2013)
2. W.W. Chen, J.F. Chen, T.J. Liang, J.R. Huang, L.C. Wei, W.Y. Ting, Implementing dynamic
quick response with high-frequency feedback control of the deformable constant on-time
control for Buck converter on-chip. IET Power Electron. 6(4), 383–391 (2013)
168 6 Ripple-Based Constant Frequency On-time Control Circuit with …

3. W.W. Chen, J.F. Chen, T.J. Liang, S.F. Hsiao, J.R. Huang, W.Y. Ting, Improved transient
response using HFFC circuit of the CCRCOT with native AVP design for voltage regulators.
IET Power Electron. 6, 1948–1955 (2013)
4. W.W. Chen, J.F. Chen, T.J. Liang, J.R. Huang, W.Y. Ting, Improved Transient Response
Using HFFC in Current-Mode CFCOT Control for Buck Converter, in Proceedings of IEEE
International Conference on Power Electronics and Drive Systems (PEDS), (2013),
pp. 546–549
5. C.J. Chen, D. Chen, C.W. Tseng, C.T. Tseng, Y.W. Chang, K.C. Wang, A novel ripple-Based
constant On-Time control with virtual inductor current ripple for buck converter with ceramic
output capacitors, in Proceedings of IEEE Applied Energy Conversion Congress and
Exposition conference, (2011), pp. 1244–1250
6. R. Redl, J. Sun, Ripple-based control of switching regulators an overview. IEEE Trans. Power
Electron. 24, 2669–2680 (2009)
7. J. Sun, Characterization and performance comparison of ripple-based control for voltage
regulator modules. IEEE Trans. Power Electron. 21, 346–353 (2006)
8. W. Huang, A new control for multi-phase buck converter with fast transient response, in
Proceedings of IEEE Applied Power Electronics Conference and Exposition conference,
(2001), pp. 273–279
9. J. Li, F.C. Lee, Modeling of V2 current-mode control, in Proceedings of IEEE Applied Power
Electronics Conference and Exposition conference, (2009), pp. 298–304
10. K.D.T. Ngo, S.K. Mishra, M. Walters, Synthetic-ripple modulator for synchronous buck
converter. Proceed. IEEE Power Electron. 3, 148–151 (2005)
11. Y.H. Lee, S.J. Wang, K.H. Chen, Quadratic differential and integration technique in V2
control buck converter with small ESR capacitor. Proc. IEEE Trans. Power Electron. 25, 829–
838 (2010)
12. M.Y. Yen, P. Mok, A constant frequency output-ripple voltage-based buck converter without
using large ESR capacitor. IEEE Transactions on Circuits and Systems 55, 748–752 (2008)
13. K.Y. Cheng, F.Yu, P. Mattavelli, F.C. Lee, Characterization and performance comparison of
digital V2-type constant on-time control for buck converters, in IEEE Control and Modeling
for Power Electronics conference, (2010), pp. 1–6
14. Richtek Tech. Corp., Comparison of DCR current sense topologies, Application Note, 2015
15. Texas Instruments Inc., TPS65311-Q1 BUCK1 controller DCR current sensing, Application
Report, 2016
16. S. Saggini, A. Zafarana, D. Zambotti, M. Ghioni, Digital autotuning system for inductor
current sensing in voltage regulation module applications. IEEE Trans. Power Electron. 23(5)
(2008)
17. Y. Yan, F.C. Lee, P. Mattavelli, Unified three-terminal switch model for current mode
controls. IEEE Trans. Power Electron. 27(9), 4060–4070 (2012)
18. L. Hua, S. Luo, Design considerations for small signal modeling of DC-DC converters using
inductor dcr current sensing under time constants mismatch conditions, in Power Electronics
Speciallists Conference, 2007, PESC’07, (2007), pp. 2182–2188
19. Y. Sun, J. Li, F.C. Lee, Modeling of digitally controlled voltage regulator modules, in Applied
Power Electronics Conference and Exposition, 2010, APEC ‘10, (2010), pp. 176–182
20. Texas Instruments Inc., “D-CAP™ Mode with all-ceramic output capacitor application,
Application Report, 2011
21. S. J. Wang, Y.H. Lee, Y.C. Lai, K.H. Chen, Quadratic differential and integration technique in
V2 control buck converter with small ESR capacitor, in Proceedings IEEE Custom Integrated
Circuits Conference, (2009), pp. 211–214
22. R. Redl, G. Reizik, Switched noise filter for the buck converter using the output ripple as
the PWM ramp, in Proceedings IEEE Applied Power Electronics Conference, (2005),
pp. 918–924
23. R. Redl, T. Schiff, A new family of enhanced ripple regulators for power-management
applications (Nuremburg, Germany, International Exhibition and Conf. Eur., 2008),
pp. 255–268
References 169

24. Richtek Tech. Corp., “3A, 18 V, 700 kHz ACOTTM synchronous step-down converter,
RT7275/RT7276 Datasheet, 2017
25. Texas Instruments Inc., TPS54325 4.5-V to 18-V, 3-A Output synchronous step down
switcher with integrated FET, TPS54325 Datasheet, 2014
26. Texas Instruments Inc., Single-phase, D-CAP™ and D-CAP2™ Controller with 2-Bit flexible
VID control, TPS51518 Datasheet, 2011
27. Richtek Tech. Corp., DCR Temperature Compensation,” Application Note, 2014
28. Richtek Tech. Corp., Multi-Phase PWM controller for CPU core power supply, RT8859 M
Datasheet, 2014
29. Richtek Tech. Corp., Dual output 3-Phase + 2-Phase PWM controller for CPU and GPU Core
power supply, RT8885A Datasheet, 2012
Chapter 7
Constant Current Ripple On-Time
Control Circuit With Native Adaptive
Voltage Positioning Design for Voltage
Regulators

7.1 Challenges for Voltage Regulators

In recent years, over one billion transistors have been integrated in one processor,
core static current has been increased from 20 to 100 A, and core voltage has been
reduced from 2 to 0.7 V [1–3]. It is a challenge to provide the large output loading
requirement for CPU application [4–8].
Single-phase VR can be widely used in low-voltage converter applications with
an output loading of up to approximately 25 A. However, power dissipation, power
stress of the components, and efficiency become an issue under a large output load
current. A suitable approach is to use multiphase VRs, as shown in Fig. 7.1. The
benefits of using multiphase VRs versus single-phase VRs and the value of mul-
tiphase VRs become evident when they are implemented.
In multiphase VRs, the phases are shifted based on the number of power trains.
For example, a three-phase converter is phase shifted by 120° (0°, 120°, and 240°)
and so on. Moreover, each phase needs to have a minimum individual off-time
limit. The interleaved control also reduces the output voltage ripple effectively.
Hence, the phases are interleaved. Interleaving effectively reduces ripple currents at
the input and output terminals. It also reduces hot spots on a PCB or a component.
In effect, two-phase VRs reduce the RMS current power dissipation in the power
MOSFET and inductors by half. Interleaving also reduces transitional losses.
The output filter requirements decrease in the implementation of multiphase VRs
due to the reduced current in the power stage for each phase. For an operation
involving 100 A of output load current, an average current of only 33 A is delivered
to each inductor when multiphase VRs are adopted. Compared with using a
single-phase approach for this 100 A scenario, the inductance and inductor size are
drastically reduced because of low average and saturation currents. The ripple
current of output load cancellation in the output-filter stage in multiphase VRs
results in a reduced ripple voltage across the output capacitor compared with the

© Springer Nature Singapore Pte Ltd. 2018 171


W.-W. Chen and J.-F. Chen, Control Techniques for Power
Converters with Integrated Circuit, Power Systems,
https://doi.org/10.1007/978-981-10-7004-4_7
172 7 Constant Current Ripple On-Time Control Circuit …

S1X LX

ILX
S2X

S12 L2

IL2
S22

VIN S11 L1 VOUT

IL1 ILOAD
S21 RCO
CO

Fig. 7.1 Control structure of multiphase VRs

situation in a single-phase VR. For these reasons, multiphase converters are pre-
ferred over single-phase ones.
Recent CPU developments have enhanced power loss reduction when the CPU
is on standby. To improve power loss, it is must reduce switching loss and con-
duction loss when the load is light. The control scheme of the adaptive on-time of
VRs can reduce switching loss and then increase light load efficiency [9–23]
because the switching frequency depends on the loading condition. If the load is
light, the switching frequency is reduced to increase efficiency. On the other hand,
the PSM mode [24–29] can be combined with the constant on-time to increase light
load efficiency because the PSM function turns off the lower-side switch when the
inductor current is equal to 0 A. It not only reduces switching loss, but also saves
system conduction loss.
The other challenge is to reduce the output capacitors, the bill of material size,
and the bill of material cost. Most of VRs have an AVP function to improve or to
solve this challenge [3, 9, 30–37]. AVP concept has been widely used in VRs of the
CPU application for energy saving in CPU power. As shown in Fig. 7.2, the
steady-state target load-line with ideal AVP for VRs, the steady-state output voltage
is lower at full-load. This reduces the power loss and heat problem of CPU in
full-load.
Output voltage waveforms when the load current goes through step changes with
ideal AVP for VRs are as shown in Fig. 7.3. The load-line is bounded by the
tolerance band indicated by the two dotted lines which is about 38 mV from Intel’s
VR12.0 specification. When the load changes, it is inevitable that output voltage
jumps but the voltage must stay within the tolerance band. If output voltage goes
7.1 Challenges for Voltage Regulators 173

Fig. 7.2 Steady-state target VOUT (V) VOUT max.


load-line with ideal AVP for VOUT typical
VRs VREF (V) VOUT min.

Slope = - RLL

Tolerance
Band

t
IOUT (A)

Fig. 7.3 Output voltage IOUT (A)


waveforms for step-load
changes with ideal AVP for IOUT
VRs
t
VOUT (V)
VOUT

above upper tolerance band, CPU may have heat problem; if output voltage goes
below lower tolerance band, computer may shut down.
Another benefit of AVP concept is the reduction of the output voltage deviation
during load transient. Figure 7.4 shows the load transient response of CPU VRs
with and without AVP. For a conventional control without AVP, output voltage is
regulated at a fixed level. On the other hand, when AVP is implemented, output
voltage stays in the voltage near upper tolerance band after load release. As can be
seen from Fig. 7.4, the voltage deviation for control with AVP is about half of the
control without AVP. This relaxes the required number of output capacitors in
parallel to reduce equivalent ESR value while increase equivalent capacitance. The
reduction of capacitor number is meaningful when considering the large percentage

Fig. 7.4 Load transient IOUT


response of CPU VRs with
and without AVP

upper limit of voltage tolerance

VOUT w/ AVP
VOUT w/o AVP
lower limit of voltage tolerance
174 7 Constant Current Ripple On-Time Control Circuit …

Fig. 7.5 Equivalent circuit of


IOUT
a VR with ideal AVP RLL

VREF
VOUT CPU

of area occupied by VRs components in computer motherboard. Besides, imple-


mentation of this concept not only reduces the power converter output capacitor
requirement but also lowers the power consumption in the microprocessor loads at
full load.
It is pointed out that to achieve ideal AVP. The constant small-signal output
impedance versus frequency is required. In addition, this output impedance must be
equal to the magnitude of load-line slope.
As can be seen from Fig. 7.5, the output voltage with ideal AVP can be
expressed by Eq. (7.1), where RLL is defined as load-line slope. It can be seen that
VRs has constant output impedance which is equal to load-line slope RLL, and CPU
is modeled by an ideal current source.

VOUT ¼ VREF  IOUT  RLL ð7:1Þ

7.2 Native Adaptive Voltage Positioning Design


for Voltage Regulators

The AVP design of the adaptive on-time control circuit for VRs has a DC offset of
the VOUT voltage by Eq. (7.2), this Equation is not the same to the Eq. (7.1), so the
AVP design of the adaptive on-time control circuit for VRs cannot achieve high
accuracy the VOUT voltage. Unfortunately, the VRs should be possessed a high
accuracy the VOUT voltage to avoid the VOUT voltage may impact the load line
specification.
   
DVOUT Ri DIL
VOUT ¼ VVID þ   IL  ð7:2Þ
2 AV 2

Native AVP design of the adaptive on-time control circuit for VRs can cancel
the DC offset of the VOUT voltage as shown in Fig. 7.6 [9, 30–41]. This control
circuit not only has adaptive on-time control circuit, but also the ability to cancel the
steady state error. The EA is an error amplifier and VCOMP1 signal is the output of
error amplifier. Two input signals of error amplifier are the feedback signal VFB and
7.2 Native Adaptive Voltage Positioning Design for Voltage Regulators 175

S13 L3

VUG3 IL3
VLG3 S23
Driver3

S12 L2

VUG2 IL2
VLG2 S22
Driver2

VIN S11 L1 VOUT


VUG1 IL1 ILOAD
VLG1 S21 RCO
Driver1
CO
R1 C1
VVID VFB
Interleaved Ri
and PSM

+
-
Control EA R2 C2

VCS VCOMP1
On-Time VTRIG
-

Generator + VCOMP2

LPF VLPF

Fig. 7.6 Native AVP design of the adaptive on-time control circuit for VRs

the VVID voltage, if the user wants to change the output voltage, and the user should
be based on VR12 VID code table setting VID bits by I2C interface [9, 30, 38, 41].
S11, S12 and S13 are the upper-side switches, S21, S22 and S23 are the lower-side
switches, L1, L2, and L3 are the output inductors, and RCO is the ESR of the output
capacitor CO. The current source ILOAD is the output load current. The feedback
resistors, R1 and R2 can be designed by load line requirement, so the output voltage
VOUT is not equal to the VVID voltage. The output voltage VOUT depends on the
output load current, if the system is operated from the light load to the heavy load
and the output voltage VOUT also should be reduced to meet the load line
requirement. The trigger signal VTRIG is the output of comparator to control on-time
generator.
The steady state waveforms of the native AVP design are as shown in Fig. 7.7.
A near-DC voltage is added with the output voltage VCOMP1 of the error amplifier.
The VCOMP2 is lower than VCOMP1 in steady state. The VOUT_NAVP is without DC
offset and it is also equal to expected value of the output voltage. The VOUT_OFS
voltage is a droop between the VOUT_NAVP voltage and the VOUT_AVP voltage. The
VCOMP_OFS voltage is a droop between the VCOMP1 voltage and VCOMP2 voltage.
The DC voltage is generated by a low pass filter subtraction of the VCS and VCOMP2
176 7 Constant Current Ripple On-Time Control Circuit …

Fig. 7.7 Steady state V


waveforms of the native AVP
design
VCOMP_OFS
VCOMP1
VCS
VCOMP2 A

t
VOUT_OFS
VOUT_AVP
VOUT_NAVP

t
VUGx

t
VTRIG

ripple voltages which are in front of the comparator. The native AVP design of the
adaptive on-time control circuit for VRs can use a low pass filter subtraction to
cancel the DC offset of the VOUT voltage, so the native AVP design can achieve the
high accuracy voltage VOUT to meet load line specification. Moreover, the corner
frequency of low pass filter should be designed much smaller than the switching
frequency. Refer to the steady-state control signals at the “A” point shown in
Fig. 7.7, and Eqs. (7.2)–(7.5) are obtained [9, 30, 38, 41].

VCS;Vally ¼ VCOMP2;Peak ð7:2Þ


 
DIL
VVID þ Ri  IL  ¼ VCOMP1 þ VLPF ð7:3Þ
2
    
DIL DVOUT
VVID þ Ri  IL  ¼ VVID þ AV  VVID  VOUT  þ VLPF
2 2
ð7:4Þ

DVOUT DIL
VLPF ¼ AV  Ri ð7:5Þ
2 2

VLPF is described in Eq. (7.5) and, by substituting (7.5) for (7.4).


Equations (7.6)–(7.8) can be obtained.
7.2 Native Adaptive Voltage Positioning Design for Voltage Regulators 177

Ri
VOUT ¼ VVID  IL ð7:6Þ
AV

Ri
RDROOP ¼ ð7:7Þ
AV

R2
AV ¼ ð7:8Þ
R1

AV is the desired error amplifier gain. Ri is the internal current sense amplifier
gain. RDROOP is the current sense resistor. This control also implements the AVP
function easily and solves a DC offset of the voltage VOUT. RDROOP should be
designed to determine the load line. It is the equivalent load line resistance as well
as the desired static output impedance.
An optimized compensation of a multiphase voltage regulator allows for the best
possible load step response of its output. A type-II compensator with one pole and
one zero is adequate for proper compensation. A prior design procedure shows how
to decide the resistive feedback components of an error amplifier gain. C1 and C2
can be calculated for compensation by Eq. (7.9) [9, 30, 38, 41]. The target is to
achieve constant resistive output impedance over the widest possible frequency
range.

Ri 1 þ s=ðR1  C1 Þ
GVC ðsÞ ¼ ð7:9Þ
RDROOP 1 þ s=ðR2  C2 Þ

7.3 On-Time Generator Circuit of the Constant Current


Ripple On-Time Control Circuit for Voltage
Regulators

On-time generator circuit of the constant current ripple on-time control circuit is as
shown in Fig. 7.8, which samples the input voltage and the VVID voltage to
determine the width of the on-time. The TON is a pin of the IC through a resistor
connected to the input voltage. It also samples the VVID voltage by Eq. (7.10). The
voltage drop between input voltage and the VVID voltage depends on the inductor
current ripple by Eq. (7.11). Therefore, even if the voltage drop between the input
voltage and the VVID voltage is changed, the inductor current ripple maintains a
constant value. Hence, the constant current ripple on-time control circuit is suitable
for different input and VVID voltages to obtain the same output voltage ripple, such
as CPU applications [9, 30–41].
178 7 Constant Current Ripple On-Time Control Circuit …

VDD

VIN RTON TON


G1
On-Time

VID
VTRIG S1 CC VREF1

Fig. 7.8 On-time generator circuit of the constant current ripple on-time control circuit

RTON
TON ¼ CC  VREF1 ð7:10Þ
VIN  VID

RTON
TON ¼ DIL ð7:11Þ
VIN  VOUT

When the resistor of RTON is changed, the width of on-time is also changed.
Switching frequency depends on the voltage drop between the input voltage and the
VVID voltage. Figure 7.9 shows the experimental results for the switching frequency
versus the output load current of the constant current ripple on-time control circuit
(VIN = 12 V, VVID = 0.7 V). Experimental results show that at an output load

Switching Frequency vs. Outupt Load


440
Switching Frequency (kHz)

400

360

320

280

RTON=150 k
240
0 10 20 30 40 50 60 70 80 90
Output Load (A)

Fig. 7.9 Experimental results of the switching frequency versus the output load current for
constant current ripple on-time control circuit
7.3 On-Time Generator Circuit of the Constant Current Ripple … 179

Output Voltage Ripple vs. Output Load


10

Output Voltage Ripple (mV)


8

RTON=150 k
0
0 10 20 30 40 50 60 70 80 90
Output Load (A)

Fig. 7.10 Experimental results of the output voltage ripple versus the output load current for
constant current ripple on-time control circuit

current of 0–90 A, when the resistor of RTON is 150 kX, the switching frequency
increases from 288 to 395 kHz.
Experimental results of the output voltage ripple versus the output load current
for constant current ripple on-time control circuit are as shown in Fig. 7.10
(VIN = 12 V, VVID = 0.7 V). Experimental results show the operation at an output
load current of 0–90 A. The inductor current ripple maintains a consistent value to
obtain the same output voltage ripple, such as 8 mV.
Experimental results of the output voltage ripple for constant current ripple
on-time control circuit are as shown in Fig. 7.11. The operation conditions are set at
0.7 V output voltage and 90 A output load current to measure the output voltage
ripple for 12 V input voltage.

Fig. 7.11 Experimental


results of the output voltage
ripple for constant current
ripple on-time control circuit
VOUT (10mV/1µs) 8mV
180 7 Constant Current Ripple On-Time Control Circuit …

7.4 Comparison of Quick Dynamic Response


and Conventional Quick Response of the On-Time
Generator Circuit

CPU load transient may occur within 1 ls and have a large output load variation.
Constant current ripple on-time control circuit also needed to add a non-linear
open-loop quick response circuit to improve transient response. Further, most
Constant current ripple on-time control circuit deliver energy from input terminal to
CPU, try to solve this issue by setting up droop-voltage thresholds to trigger another
open-loop regulation mechanism, such as triggering another on-time generator or
increasing the duty of its on-time generator. However, this kind of design has two
major drawbacks. First, the threshold is discrete, which means it may improve
transient response over a specific threshold. Second, the threshold is fixed, which
cannot meet the verity of loading conditions. Moreover, if the threshold can be set
by external components, it suffers another drawback, extra pins, which increase cost
and reduce the flexibility of board design.
Recently, two types of improved transient response have been used, namely, the
internal fixed quick response by IC design, and the external setting by component
or voltage source. The internal fixed quick response by IC design usually uses VFB
to determine and trigger the quick response function, which can provide longer
on-time width to the driver circuit. The main advantage of this process is that no
extra pin in the IC is needed to achieve this function. However, this generator
circuit of the quick response function is not convenient to use in CPU applications
because it cannot be designed to control the on-time width during different load
conditions. On the other hand, the hysteresis of the comparator is hard to design,
especially for very fast load transient like CPU applications. If the hysteresis of the
comparator is too small, it can induce the system to trigger a longer on-time width
erroneously, thereby causing a high ripple of VOUT. Conversely, if the hysteresis of
the comparator is too large, it can render the quick response function unavailable
[9, 30–41].
External setting quick response of the on-time generator circuit for constant
current ripple on-time control circuit is as shown in Fig. 7.12. The operational
principle of the external setting quick response requires the addition of a voltage
source VQRSET to determine the width of QR-time. VQRSET is connected to the
“QRSET” pin. If the QR-time needs a longer on-time width, the user should design
VQRSET to be larger than or equal VREF1 [9, 30, 38, 41].
The QR trigger circuit samples the VOUT signal and uses the low-pass filter to
cause the VOUT signal delay and a DC source of VQRTH. At a steady state, the drop
of VOUT cannot trigger the QR-time. When this abrupt voltage drop is lower than
the QR threshold level, the QR trigger circuit generates a low-level signal that turns
off the switch of S2 as shown in Fig. 7.13. The frequency of the low-pass filter for
RLOW and CLOW should be designed to be much smaller than the switching fre-
quency which can avoid this system to have a failure in operation with QR-time.
The VQRTH signal can be designed in IC or set by the user for “QRSET” pin.
7.4 Comparison of Quick Dynamic Response and Conventional Quick … 181

VIN RTON TON G1


VDD
On-Time

VTRIG S1 CC1 VREF1

G2
VDD
QR-Time
VID

VQRTH
VOUT VA VQR_TRIG CC2
VB S2
QRSET
RLOW

QR trigger VQRSET
CLOW
circuit

Fig. 7.12 External setting quick response of the on-time generator circuit for constant current
ripple on-time control circuit

Fig. 7.13 Control signals for


QR trigger circuit of external VA (VOUT+VQRTH)
setting quick response (VOUT)

VB (VOUT)

VQR_TRIG (VDD)

(0)

ILOAD
(Heavy Load)

(No Load)

The main advantage of the external setting quick response can depend on the
load conditions to design the VQRSET value generating a suitable QR-time in this
system. However, this method requires two extra IC pins to achieve the quick
response function.
182 7 Constant Current Ripple On-Time Control Circuit …

VIN

RTON
I1

ISUM G1
TON VDD
On-Time

Rq I2

VID VTRIG S1 CC VREF1


Cq

VOUT

Fig. 7.14 Quick dynamic response of on-time generator circuit for constant current ripple on-time
control circuit

Quick dynamic response of the constant current ripple on-time control circuit
with native AVP design for voltage regulators is proposed as shown in Fig. 7.14. It
is an external setting device that uses these components design, not a voltage source
to set up. This quick dynamic response does not require any extra pin to achieve the
quick response function of ICs. The system consists of resistor Rq and capacitor Cq
in series that are connected between the VOUT and TON pin. It operates under the
principle that the high-pass filter filters high-frequency signals to pass from the
VOUT to the TON pin by Eq. (7.12). However, the frequency of the high-pass filter
must be larger than or equal the switching frequency so the steady state operation of
the system is not affected. Finally, the user may design the resistor Rq and the
capacitor Cq based on the worst-case operation (maximum load step) [9, 30, 38, 41].

1
FRC ¼  FS ð7:12Þ
2p  Cq  Rq

When light load quickly transforms quickly into a heavy load, VOUT drops
momentarily through the coupling by the capacitor Cq, which allows the resistor Rq
to cause a voltage drop. The voltage drop of Rq forms a current of I2, so the Rq
needs to be designed first because the current of I2 can change the width of on-time
directly as shown in Fig. 7.15. However, a longer width of on-time makes the
converter to deliver more energy from the input terminal to the output loading. If Rq
has changed small, the width of on-time becomes longer. On the other hand, the
capacitor of Cq should be designed with the frequency of the high-pass filter. The
calculation of the Laplace transform formula is shown by Eq. (7.13).
7.4 Comparison of Quick Dynamic Response and Conventional Quick … 183

Fig. 7.15 Control signals for


quick dynamic response of VOUT (VID)
on-time generator circuit for
constant current ripple (VID-RDROOP*IL)
on-time control circuit
I2 (0)

ISUM (I1)

ILOAD (Heavy Load)

(No Load)
VUGX (VDD)

(0)

C1
TON ðsÞ ¼   VREF1
G1  ðVIN  VIDÞ=RTON þ s  Cq  VOUT  VID 1 þ s  Cq  Rq
ð7:13Þ

In the multiphase operation, the quick dynamic response not only increases
longer on-time width directly, but also controls the driver in each phase
synchronously.
The advantages of the quick dynamic response are the follows:
1. It clearly generates a longer width of on-time that is proportional to the output
voltage drop.
2. Adaption to the width of on-time depends on the load conditions of the system.
3. It only uses one generator circuit; thus, it is very convenient to design and apply.
4. It does not require an extra pin to achieve the quick dynamic response function.

7.5 Experimental Results

To understand the feasibility and the performance of the quick dynamic response of
the constant current ripple on-time control circuit with native AVP design for voltage
regulators, experimental results with multiphase operation are shown to prove that it
is viable and useful. The operation conditions are as follows [9, 30, 38, 41]:
(1) Input voltage (VIN): 12 V
(2) Output voltage (VVID): 0.7 V
184 7 Constant Current Ripple On-Time Control Circuit …

(3) Maximum output load current (ILOAD): 35–75 A @ 75 A/ls


(4) Switching frequency (FS): 288–395 kHz
(5) Upper-side MOSFET (S11, S12, S13): IPS09N03LA * 3
(6) Lower-side MOSFET (S21, S22, S23): IPS06N03LA * 6
(7) Compensation circuit: R1 = 10 kX, R2 = 47 kX, C1 = 80 pF, C2 = 100 pF
(8) Load line (RDROOP): 2.25 mX
(9) Current Sense Amplifier Gain (Ri): 10
(10) Main inductor (L1, L2, L3): 360 nH * 3
(11) Output capacitors (CO): COS-CAP = 560 lF/2.5 V (RCO: 7 mX) * 3,
CMLCC-CAP = 22 lF/6.3 V (RCO: 3 mX) * 22
(12) Quick dynamic response (Fig. 7.8): G1 = 147.54 m (A/A); RTON = 150 kX;
CC = 3 pF; Rq = 330 X; Cq = 560 pF; VREF1 = 1.2 V
(13) External setting quick response (Fig. 7.6): RLOW = 600 kX;
CLOW = 2 pF; G1 = 147.54 m (A/A); RTON = 150 kX; CC = 3 pF;
VREF1 = 1.2 V; VQRSET = 1.2 V; VQRTH = 35 mV
Figure 7.16 shows the chip layout of the control IC, in which the on-time
generator and loop control are both occupied. The area is marked with a die size of
3040 lm  2300 lm.
Experimental results at the droop with the quick dynamic response (Fig. 7.14)
and without quick response (Fig. 7.8) are as shown in Fig. 7.17. The blue

Protection and sequence


TON Gen
Diagnostic

TON Gen
Loop
2300 µm

Control

Loop
Control

3040 µm

Fig. 7.16 Chip layout of the control IC


7.5 Experimental Results 185

Fig. 7.17 Comparison of the


experimental results at the (1)
droop with quick dynamic 88mV
response and without quick VOUT(50mV/10µs)
response
(2)
109mV
VOUT(50mV/10µs)

ILOAD(A/10µs) (3)
75A

35A
(2) Without Quick Response
(1) Quick Dynamic Response

waveform has without quick response for the output voltage and the red waveform
has the quick dynamic response for the output voltage. The output voltage signals
are based on the same measured load current, such as the black waveform, which
uses a voltage transient test (VTT) tool to achieve load transient [9, 30, 38, 41].
The VTT tool is created by Intel Corporation to simulate a CPU behavior, such as
dynamic VID variation and dynamic load transient. The VTT tool is used to
generate the maximum current slew rate of 300 A/ls for load transient.
VOUT peak to peak value at the load transient with the quick dynamic response is
88 mV and VOUT peak to peak value at the load transient without quick response is
109 mV. Hence, VOUT with the quick dynamic response is not only lower by
21 mV than without quick response, but also has a lower voltage to meet the load
line specification as shown by the black dot line. The load line specification has
90 mV from VR12 specification definition and it is also equal to the RDROOP
2.25 mX to multiply the output loading step 40 A.
Experimental results of the output voltage and control signals at the droop
without quick response are as shown in Fig. 7.18. The blue waveform is without

Fig. 7.18 Experimental


results of the output voltage
and control signals at the
droop without quick response VOUT(50mV/2µs)

VUG1 (20V/2µs)

VUG2 (20V/2µs)

VUG3 (20V/2µs)

(1) Without Quick Response of VOUT


(2),(3),(4) Without Quick Response of Each VUGX
186 7 Constant Current Ripple On-Time Control Circuit …

quick response for the output voltage, the green waveform is without quick
response for the VUG1 signal, the orange waveform is without quick response for the
VUG2 signal, the pink waveform is without quick response for the VUG3 signal.
These control signals have the same width of on-time and a sequencing control. On
the other hand, these control signals cannot drive each phase synchronously to
deliver more energy from the input terminal to the output loading.
Experimental results of the output voltage and control signals at the droop with
quick dynamic response are as shown in Fig. 7.19. The red waveform is that with
quick dynamic response for the output voltage signal, the green waveform is that
with quick dynamic response for the VUG1 signal, the orange waveform is that with
quick dynamic response for the VUG2 signal, the pink waveform is that with quick
dynamic response for the VUG3 signal. These control signals with quick dynamic
response are obviously longer than those without quick response during the output
loading, from light load to heavy load. On the other hand, these control signals can
drive all phases synchronously to deliver more energy from the input terminal to the
output loading, the output voltage has increased after quick dynamic response
triggered.
Simulation results of the output voltage and control signals at the droop with
quick dynamic response are as shown in Fig. 7.20. The red waveform is that with
quick dynamic response for the output voltage signal, the green waveform is that
with quick dynamic response for the VUG1 signal, the orange waveform is that with
quick dynamic response for the VUG2 signal, the pink waveform is that with quick
dynamic response for the VUG3 signal. These control signals with quick dynamic
response are obviously longer than those without quick response during the output
loading, from light load to heavy load. On the other hand, these control signals can
drive all phases synchronously to deliver more energy from the input terminal to the
output loading to prevent the output voltage from dropping significantly. Thus,
SIMPLIS simulation and experimental results are matched.

Fig. 7.19 Experimental


results of the output voltage
and control signals at the VOUT (50mV/2µs)
droop with quick dynamic
response
VUG1 (20V/2µs)

VUG2 (20V/2µs)

VUG3 (20V/2µs)

(1) Quick Dynamic Response of VOUT


(2),(3),(4) Quick Dynamic Response of Each VUGX
7.5 Experimental Results 187

84mV VOUT (10mV/5us)

VUG1 (20V/2µs) Quick Dynamic Response

VUG2 (20V/2µs)

VUG3 (20V/2µs)

ILOAD (5A/2µs)

Fig. 7.20 Simulation results of the output voltage and control signals at the droop with quick
dynamic response

To further understand the advantage and the superiority of the quick dynamic
response, the plans to base identical operation conditions by comparing the external
setting quick response.
Experimental results at the droop with quick dynamic response (Fig. 7.14) and
external setting quick response (Fig. 7.12) are as shown in Fig. 7.21. The blue
waveform is that with external setting quick response for the output voltage, the red
waveform is that with quick dynamic response for the output voltage. The output
voltage signals are based on the same measured load current, such as the black
waveform used as VTT tool to achieve load transient. VOUT peak to peak value at
the droop with quick dynamic response is 88 mV and VOUT peak to peak value at
the droop with external setting quick response is 96 mV. Thus, VOUT with quick
dynamic response and external setting quick response are close. Quick dynamic
response and external setting quick response are useful to prevent the VOUT from
dropping significantly, but the quick dynamic response does not require an extra pin
to achieve the quick response function.
188 7 Constant Current Ripple On-Time Control Circuit …

Fig. 7.21 Comparison of the


experimental results at the (1)
droop with quick dynamic
88mV VOUT (50mV/10µs)
response and external setting
quick response
(2)
96mV VOUT (50mV/10µs)

ILOAD (A/10µs) (3)


75A

35A
(2) External Setting Quick Response
(1) Quick Dynamic Response

7.6 Summary

This chapter proposes to achieve quick dynamic response of the constant current
ripple on-time control circuit with native AVP design for voltage regulators. The
concept uses the capacitor and resistor in series to filter VOUT at load transient to
change the width of on-time dynamically to prevent the VOUT from dropping
significantly.
Both experimental and simulation results confirm that the proposed the quick
dynamic response of the constant current ripple on-time control circuit with native
AVP design for voltage regulators can significantly improve transient response.
Moreover, the proposed quick dynamic response has very simple structure and
design with native AVP design for voltage regulators.

References

1. E. Stanford, Power technology roadmap for microprocessor voltage regulators, in


Presentation at PSMA, Feb 2003
2. 2001 technology roadmap for semiconductors, available: http://www.intel.com/research/
silicon/AlanAllanIEEEComputer0102.pdf
3. K. Yao, Y. Ren, J. Sun, K. Lee, M. Xu, J. Zhou, F.C. Lee, Adaptive voltage position design
for voltage regulators, in Proceedings of IEEE Applied Power Electronics Conference and
Exposition Conference (2004), pp. 272–278
4. H. Mao, L. Yao, C. Wang, I. Batarseh, Analysis of Inductor current sharing in nonisolated and
isolated Multiphase DC–DC converters. IEEE Trans. Ind. Electron. 54(6), 3379–3388 (2007)
5. P.-W. Lee, Y.-S. Lee, D.K.W. Cheng, X.-C. Liu, Steady-state analysis of an interleaved boost
converter with coupled inductors. IEEE Trans. Ind. Electron. 47(4), 787–795 (2000)
6. H.N. Nagaraja, D. Kastha, A. Patra, Design principles of a symmetrically coupled Inductor
structure for Multiphase synchronous Buck converters. IEEE Trans. Ind. Electron. 58(3),
988–997 (2011)
7. L.-P. Wong, D.K.-W. Cheng, M.H.L. Chow, Y.-S. Lee, Interleaved three-phase forward
converter using integrated transformer. IEEE Trans. Ind. Electron. 52(5), 1246–1260 (2005)
References 189

8. J. Abu-Qahouq, H. Mao, I. Batarseh, Multiphase voltage-mode hysteretic controlled DC–DC


converter with novel current sharing. IEEE Trans. Power Electron. 19(6), 1397–1407 (2004)
9. W.W. Chen, J.F. Chen, T.J. Liang, S.F. Hsiao, J.R. Huang, W.Y. Ting, Improved transient
response using HFFC circuit of the CCRCOT with native AVP design for voltage regulators.
IET Power Electron. 6, 1948–1955 (2013)
10. C. Ni, T. Tetsuo, Adaptive constant on-time (D-CAP™) control study in notebook
applications, in Texas Instruments, Application report SLVA281B, July 2007
11. R. Redl, J. Sun, Ripple-based control of switching regulators an overview. IEEE Trans. Power
Electron. 24, 2669–2680 (2009)
12. J. Li, Current-mode control: modeling and its digital application, Ph.D. thesis, Virginia
Polytechnic Institute and State University, 2009
13. J. Sun, Characterization and performance comparison of ripple-based control for voltage
regulator modules. IEEE Trans. Power Electron. 21, 346–353 (2006)
14. W. Huang, A new control for multi-Phase buck converter with fast transient response, in
Proceedings of IEEE Applied Power Electronics Conference and Exposition Conference
(2001), pp. 273–279
15. J. Li, F.C. Lee, Modeling of V2 current-mode control, in Proceedings of IEEE Applied Power
Electronics Conference and Exposition Conference (2009), pp. 298–304
16. S.J. Wang, Y.H. Lee, Y.C. Lai, K.H. Chen, Quadratic differential and integration technique in
V2 control buck converter with small ESR capacitor, in Proceedings of IEEE Custom
Integrated Circuits Conference (2009), pp. 211–214
17. R. Redl, G. Reizik, Switched noise filter for the buck converter using the output ripple as the
PWM ramp, in Proceedings of IEEE Application Power Electronics Conference (2005),
pp. 918–924
18. J. Li, F.C. Lee, New Modeling Approach and Equivalent Circuit Representation for
Current-Mode Control. IEEE Trans. Power Electron. 25, 1218–1230 (2010)
19. Y.H. Lee, S.J. Wang, K.H. Chen, Quadratic differential and integration technique in V2
control buck converter with small ESR capacitor. IEEE Trans. Power Electron. 25, 829–838
(2010)
20. M.Y. Yen, P. Mok, A constant frequency output-ripple voltage-based buck converter without
using large ESR capacitor. IEEE Trans. Circuits Syst. I 55, 748–752 (2008)
21. K.Y. Cheng, F. Yu, P. Mattavelli, F.C. Lee, Characterization and performance comparison of
digital V2-type constant on-time control for buck converters, in IEEE Control and Modeling
for Power Electronics Conference, June 2010, pp. 1–6
22. A.V. Petershevs, S.R. Sanders, Digital multimode buck converter control with
loss-minimizing synchronous rectifier adaptation. IEEE Trans. Power Electron. 21(6),
1588–1599 (2006)
23. J. Wang, B. Bao, G. Zhou, W. Hu, Dynamical effects of equivalent series resistance of output
capacitor in constant on-time controlled Buck converter. IEEE Trans. Ind. Electron. 58(12),
5406–5410 (2011)
24. S. Angkititrakul, H. Hu, Design and analysis of buck converter with pulse-skipping
modulation, in Proceedings of IEEE Power Electronics Specialists Conference (2008),
pp. 1151–1156
25. X. Zhou, M. Donati, L. Amoroso, F.C. Lee, Improved light-load efficiency for synchronous
rectifier voltage regulator module. IEEE Trans. Power Electron. 15(5), 826–834 (2000)
26. C.L. Chen, W.L. Hsieh, W.J. Lai, K.H. Chen, C.S. Wang, A new PWM/PFM control
technique for improving efficiency over wide load range, in Proceedings of IEEE
International Conference on Electronics, Circuits and Systems, Aug 2008, pp. 962–965
27. S. Kapat, S. Banerjee, A. Patra, Discontinuous map analysis of a DC-DC converter governed
by pulse skipping modulation. IEEE Trans. Circuits Syst. I Part I 57(7), 1793–1801 (2010)
28. C.A. Yeh, Y.S. Lai, Digital pulsewidth modulation technique for a synchronous Buck DC/DC
converter to reduce switching frequency. IEEE Trans. Ind. Electron. 59(1), 550–561 (2012)
29. J. Wang, J. Xu, B. Bao, Analysis of pulse bursting phenomenon in Constant-On-
Time-Controlled Buck converter. IEEE Trans. Ind. Electron. 58(12), 5406–5410 (2011)
190 7 Constant Current Ripple On-Time Control Circuit …

30. J.R. Huang, C.H. Wang, C.J. Lee, K.L. Tseng, D. Chen, Native AVP control method for
constant output impedance of DC power converters, in Proceedings of IEEE Power
Electronics Specialists Conference (2007), pp. 2023–2028
31. A. Waizman, C.Y. Chung, Resonant free power network design using extended adaptive
voltage positioning (EAVP) methodology. IEEE Trans. Adv. Packag. 24, 236–244 (2001)
32. P.L. Wong, Performance improvements of multi-channel interleaving voltage regulator
modules with integrated coupling inductors, Dissertation of Virginia Polytechnic Institute and
State University, Mar 2001
33. M. Lee, D. Chen, K. Huang, E. Tseng, B. Tai, Compensator Design for Adaptive Voltage
Position (AVP) for Multiphase VRMs, in Proceedings of IEEE Power Electronics Specialists
Conference (2006)
34. K. Yao, Y. Meng, P. Xu, F.C. Lee, Design considerations for VRM transient response based
on the output impedance, in Proceedings of IEEE Applied Power Electronics Conference and
Exposition Conference (2002), pp. 14–20
35. R. Redl, N.O. Sokal, Near-optimum dynamic regulation of DC-DC converters using
feed-forward of output current and input voltage with current-mode control. IEEE Trans.
Power Electron. 1, 181–192 (1986)
36. S.K. Mishra, Design-oriented analysis of modern active droop controlled power supplies.
IEEE Trans. Ind. Electron. 56(9), 3704–3708 (2009)
37. J.A.A. Qahouq, V. Arikatla, Power converter with digital sensorless adaptive voltage
positioning control scheme. IEEE Trans. Ind. Electron. 58(9), 4105–4116 (2010)
38. Richtek Technology Corporation, Dual output 3-Phase + 2-Phase PWM controller for CPU
and GPU Core power supply, RT8885A Datasheet, 2012
39. K.P. Liu, K.C. Wang, L.P. Tai, C.S. Cheng, Control circuit and method for a constant on-time
PWM switching converter, U. S. Patent 7 834 606 B2, 2 Aug 2007
40. K Huang, Spring modulation with fast load-transient response for a voltage regulator,
U. S. Patent 7 247 182 B2, 26 Aug 2005
41. Richtek Technology Corporation, Multi-Phase PWM controller for CPU core power supply,
RT8859M Datasheet, 2014

You might also like