Content uploaded by Philip Mok
Author content
All content in this area was uploaded by Philip Mok on May 15, 2014
Content may be subject to copyright.
IEEE TRANSACTIONS ON SOLID-STATE CIRCUITS, VOL. 35, NO. 2, FEBRUARY 2000 221
Three-Stage Large Capacitive Load Amplifier with
Damping-Factor-Control Frequency Compensation
Ka Nang Leung, Philip K. T. Mok, Member, IEEE, Wing-Hung Ki, Member, IEEE, and
Johnny K. O. Sin, Senior Member, IEEE
Abstract—A novel damping-factor-control frequency compen-
sation (DFCFC) technique is presented in this paper with detailed
theoretical analysis. This compensation technique improves fre-
quency response, transient response, and power supply rejection
for amplifiers, especially when driving large capacitive loads.
Moreover, the required compensation capacitors are small and
can be easily integrated in commercial CMOS process.
Amplifiers using DFCFC and nested Miller compensation
(NMC) driving two capacitive loads, 100 and 1000 pF, were
fabricated using a 0.8-
m CMOS process with 0.72 V and
0.75 V. For the DFCFC amplifier driving a 1000-pF load,
a 1-MHz gain-bandwidth product, 51
phase margin, 0.33-V/ s
slew rate, 3.54-
s settling time, and 426- W power consumption
are obtained with integrated compensation capacitors. Compared
to the NMC amplifier, the frequency and transient responses of
the DFCFC amplifier are improved by one order of magnitude
with insignificant increase on the power consumption.
Index Terms—Damping factor, frequency compensation, large
capacitive load, multistage amplifier.
I. INTRODUCTION
W
ITH the rapid decrease in the supply voltage in VLSI,
more circuit designers are aware of the importance of
low-voltage multistage amplifiers. However, all multistage
amplifiers suffer close-loop stability problems due to their
multiple-pole nature. Therefore, many frequency compensation
topologies have been proposed [1]–[6].
As mentioned by Eschauzier et al. [1] and Huijsing et al.
[2], multistage amplifiers suffer bandwidth reduction, and a
single-stage amplifier is optimum on bandwidth. Compared
to a single-stage amplifier, the gain-bandwidth products of a
two-stage simple Miller compensated (SMC) amplifier and a
three-stage nested Miller compensated (NMC) amplifier are
reduced tohalfand onequarter, respectively [1], [2]. As a result,
other advanced topologies, such as multipath NMC (MNMC),
hybrid NMC (HNMC), multipath HNMC (MHNMC), and
nested Gm-C compensation (NGCC), were developed to
overcome the bandwidth reduction problem [1]–[5]. For these
topologies, which are based on NMC, pole–zero cancellation
and feed-forward technique are used to extend the bandwidth.
However, from the findings shown in Fig. 1, the bandwidth
improvements by the above-mentioned topologies over NMC
Manuscript received July 2, 1999; revised September 30, 1999. This
work was supported by the RGC Competitive Earmarked Research Grant
HKUST6007/97E, Hong Kong SAR Government.
TheauthorsarewiththeDepartmentofElectricaland ElectronicEngineering,
The Hong Kong University of Science and Technology, Clear Water Bay, Hong
Kong, China.
Publisher Item Identifier S 0018-9200(00)00938-0.
are not sufficient, and the provided bandwidths are narrower
than the bandwidth of a single-stage amplifier. In addition,
tradeoffsbetween bandwidth and other amplifier characteristics
such as settling time and power consumption are required. In
particular, higher power consumption is needed to increase
the bandwidth when the amplifier is required to drive a large
capacitive load such as the error amplifier in a linear regulator.
Furthermore, both small-signal and settling behavior should be
improved at the same time in order to obtain a fast amplifier
[7]. Nevertheless, the above-mentioned topologies are mainly
concerned with the improvement of the frequency response.
Another issue to be considered is the optimum number of
gain stages. Generally, frequency compensation techniques for
three-stage amplifiers are adequate for practical purposes since
three-stage amplifiers maintaina good compromise between the
voltagegain
100 dB) andpowerconsumption.Anyextragain
stage complicates the circuit structure and increases the com-
plexity of the frequency compensation.
With respect to the above problems, a novel damping-factor-
control frequency compensation (DFCFC) technique [8] is pre-
sented in this paper. As illustrated in Fig. 1 and shown later in
the paper, this topology substantially increases the bandwidth
of a three-stage amplifier, especially when driving large capac-
itive loads. The improvement can be as large as one order of
magnitude.Furthermore, aDFCFCamplifieris able toprovidea
wider bandwidth than a single-stage amplifier with significantly
improved transient response and power supply rejection ratio.
To achieve the above enhancements, the required values of the
compensation capacitors are small, and there is a small increase
on the power consumption and circuit complexity.
In the next section, a brief review on NMC is given. The
purpose of the review is to take NMC as a common reference
when comparing DFCFC to all published topologies. DFCFC
is introduced in Section III. The transfer function, stability cri-
teria,transientresponse, and power supply rejection ratioaread-
dressed. In Sections IV and V, the implementation of the struc-
ture and experimental results are presented. Finally, the conclu-
sion of this paper is given in Section VI.
II. B
RIEF REVIEW ON NMC
The structure and equivalent circuit of a three-stage amplifier
using NMC is shown in Fig. 2. The transconductance, output
resistance, and lumped output parasitic capacitance of the gain
stagesarenotatedby
,and ,respectively.
and are the compensation capacitors, and is the
loading capacitance. It should be noted that in order to obtain
222 IEEE TRANSACTIONS ON SOLID-STATE CIRCUITS, VOL. 35, NO. 2, FEBRUARY 2000
Fig. 1. Bandwidth comparison of different frequency compensation topologies (take NMC as the reference).
(a)
(b)
Fig. 2. (a) Structure of a three-stage NMC amplifier. (b) Equivalent
small-signal circuit of the three-stage NMC amplifier.
negative capacitive feedback loops by and the gains
of the second stage and output stage are positive and negative,
respectively.
As there are three high impedance nodes in the structure,
there are three poles. Solving the equivalent circuit with the fol-
lowing assumptions: 1)
, , and and
and 2) the small-signal transfer function,
is calculated as
(1)
where
isthedcgain, and
is the dominant pole. To stabilize
the NMC amplifier, the NMC amplifier should have third-order
Butterworth frequency response [9] in unity-feedback configu-
ration, and
and should obey the following dimension
conditions [1]–[3]:
(2)
(3)
From the second-order function at the denominator of (1), the
positions of the second and third pole depend on
and ,
and the damping factor of the second-order function is con-
trolled by
. From (1) and (3), the nondominant poles de-
pends on
and thus depends on the loading capacitance .
This is not preferable as the loading capacitance will affect the
location of the nondominant poles and therefore affects the sta-
bility of the amplifier. When driving a large capacitive load, a
larger
is required, and the nondominant poles will locate
at rather low frequencies. Therefore, the bandwidth of an NMC
amplifier is poor.
Substituting the dimension conditions of
and
into (1), the second and third pole will form a complex pole
with damping factor of
. The gain-bandwidth product
and phase margin are [1]
(4)
(5)
From the above results, the gain-bandwidth product decreases
as the loading capacitance increases. The only method to in-
LEUNG et al.: THREE-STAGE LARGE CAPACITIVE LOAD AMPLIFIER WITH DFCFC 223
crease the bandwidth of an NMC amplifier is to enlarge
by increasing the quiescent current and size of the transistors
of the output gain stage. Moreover, from the design point of
view, NMC is not suitable for low-power design as the previ-
ouslystated assumption,
,may not be valid.
Although small bias current and small transistor size can reduce
such that is smaller than , the same does not apply
to
. This is due to the small bias current at the input stage
which reduces the slew rate of the amplifier [10]–[15] and the
small size of the input differential pair, which introduces a large
offset voltage [15]. Thus, NMC may not be suitable for ampli-
fiers driving large capacitive loads and in low-power designs.
III. DFCFC
In this section, detailed analysis on the transfer function, sta-
bility criteria, slew rate, settling time, and power supply rejec-
tion ratio of the proposed DFCFC structure are discussed.
A. Structure of DFCFC
From the discussion in the previous section, the poor band-
width of an NMC amplifier is mainly due to the presence of
, which is part of capacitive load to the amplifier [1], [6].
Byeliminating
, the capacitiveload at theoutput is reduced.
The nondominant poles will then relocate at higher frequencies,
and the bandwidth of the NMC amplifier can be extended. Nev-
ertheless, the two nondominant poles form a complex pole. The
damping factor of the complex pole is very small. Moreover,
there is no control of the damping factor as
is absent. As
shown in Fig. 3, the small damping factor causes a frequency
‘peak’ to appear near the unity-gain frequency of the amplifier,
and this prohibits stable close-loop operation.
In order to obtain a larger bandwidth and stabilize the am-
plifier at the same time, DFCFC is used. The structure and the
equivalent circuit are shown in Fig. 4. There are two additional
building blocks: 1) the feed-forward transconductance stage
(FTS) and 2) the damping-factor-control (DFC) block. The
function of the FTS is to implement a push–pull output stage
to improve the slewing performance. The DFC block is used to
control the damping factor of the nondominant complex pole to
make the amplifier stable. The resultant open-loop frequency
response of the DFCFC amplifier is shown in Fig. 3. The
DFCFC bandwidth is wider than that of the NMC amplifier.
For the details of the two additional blocks, the DFC block is
simply a negative gain stage with transconductance
, output
Fig. 3. Bode plot of the frequency responses.
resistance , and lumped parasitic capacitance The dc
gain of this stage must be greater than one to validate the math-
ematical derivation of the transfer function shown in next sec-
tion.TheFTS block is merelyatransistorwith transconductance
. The input signal of the FTS is from the output of the
first stage
in Fig. 4). As will be explained in detail in the
following sections, the two additional building blocks signifi-
cantly improve the bandwidth and transient response.
B. Transfer Function
To analyze the stability of the DFCFC amplifier, the small-
signal transfer function should be investigated. Solving the cir-
cuit network shown in Fig. 4 by setting
to sim-
plify the expression and with the following assumptions. 1) The
dc gain of the DFC block is greater than one; 2)
and
are smaller than the compensation capacitances, loading capac-
itance, and
( is generally larger than other parasitic
capacitances as it depends on the size of the transistor of the
output stage), the transfer function is given by (6), shown at the
bottom of the page, where
is the
dc gain, and
is the dom-
inant pole. It should be noted that the effect of
is can-
celed in the transfer function, so it is not always necessary to
set
. The above transfer function holds true as long
as
.If is small, should be set to equal
; otherwise, set to further optimize
the size of the compensation capacitors.
From (6), shown at the bottom of the page, it is obvious that
the damping factor and location of the complex pole can be con-
trolled by an optimum value of
, and the gain-bandwidth
(6)
224 IEEE TRANSACTIONS ON SOLID-STATE CIRCUITS, VOL. 35, NO. 2, FEBRUARY 2000
(a)
(b)
Fig. 4. (a) Structure of a DFCFC amplifier. (b) Equivalent small-signal circuit of the DFCFC amplifier.
product is controlled by . Moreover, thenondominant poles
locate at higher frequencies as the second-order function de-
pends on the parasitic capacitance
instead of , which
is loading capacitance dependent in NMC topology.
C. Stability Criteria, Gain-Bandwidth Product, and Phase
Margin
From the previous section, it is known that the stability of the
amplifier can be controlled by appropriate values of
and
. In this section, the dimension conditions are calculated.
We model the transfer function with ideal function first and
then apply the results. To begin, assuming that the effect of
the zeros is negligible and considering that the poles of the
DFCFC amplifier in unity-feedback configuration have third-
order Butterworth frequency response [9], the transfer func-
tion in unity-gain feedback
with cut-off frequency
is given by
(7)
To obtain (7), the open-loop transfer function should be in the
following format:
(8)
(9)
Considering a typical second-order function
with damping
factor
and corner frequency , is given by
(10)
By comparing the denominator of (9) with (10), it is proven that
the damping factor of the nondominant complex pole in open
loop is
if the amplifier has third-order Butterworth fre-
quencyresponse in unity-feedbackconfiguration.Therefore,the
magnitude plot of the amplifier in open loop has no frequency
“peak.” The position of the nondominant complex pole
LEUNG et al.: THREE-STAGE LARGE CAPACITIVE LOAD AMPLIFIER WITH DFCFC 225
is obtained by solving the second-order function in (8) and is
given by
(11)
The gain-bandwidth product
of the amplifier is
given by
(12)
and the phase margin
[16] is expressed by
(13)
By applying the above modeling on DFCFC, three equations
are obtained by comparing the coefficients of the denominator
of (8) with those of (6).
(14)
(15)
(16)
By substituting (15) into (16) to eliminate
, a quadratic equa-
tion of
is derived.
(17)
By solving the above equation and rejecting the negative solu-
tion of
, the dimension condition of in DFCFC is given
by
(18)
Forthedimension conditionof
,it canbe obtained byfirstly
substituting (15) into (14) to acquire
(19)
and then substituting (18) into (19) to obtain
(20)
Since it is preferable to have the same output current capability
for both the p- and the n-transistor of the output stage, the sizes
ofthep-andn-transistorareusedin the ratio of 3 to 1to compen-
sate for the difference in the mobilities of the carriers. Thus, it is
reasonable to set
, and (18) and (20) are rewritten
as
(21)
(22)
where
(23)
Since the dimension conditions of
and depend on the
ratios of transconductances and capacitances, the stability of the
DFCFC amplifier is less sensitive to global variations of circuit
parameters.
From(23),
islargerforalarger to ratio,andthevalue
canbemuchgreaterthanoneinsomecases.Therequired
isre-
ducedby
timeswhencomparedto(2)forNMC.Inaddition,since
theproductof
and in(21)isadecreasingfunctionwith
,therequired issmallandonlyasmallamountofpoweris
dissipatedintheDFCblock.
Asthezerosoftheamplifierdependon
(whichissmallas
provenpreviously)and
,theyareathigherfrequenciesthanthe
polesoftheamplifierandtheeffectfromthezeroscanbeneglected.
Therefore,thepreviouslystatedassumptionisvalid,andthepres-
enceofzerosdoesnotinvalidatethestabilitycriteria.
From (12) to (14), the gain-bandwidth product and phase
margin of the DFCFC amplifier are given by
(24)
(25)
It can be shown from (24) that the bandwidth of a DFCFC am-
plifier is
times that of an NMC amplifier. The improvement
by DFCFC is greater for a larger
(i.e., a larger loading ca-
pacitance). Referring to Fig. 1, the bandwidth of the DFCFC
amplifier may be even larger than a single-stage amplifier when
is larger than 4. Also, there is no degradation on the phase
margin when using DFCFC. If the required phase margin is less
than 60
in some applications, a smaller can be used to de-
crease the phase margin and further increase the bandwidth of
the amplifier.
226 IEEE TRANSACTIONS ON SOLID-STATE CIRCUITS, VOL. 35, NO. 2, FEBRUARY 2000
D. Slew Rate
After the analysis on AC behavior, the transient behavior is
studied in this section and the next section. Assuming that the
output stage is of push–pull or class-AB type, the slew rate is
not limited by the output stage. Thus, the slew rate of a multi-
stage amplifier depends on two factors. One is the value of the
compensation capacitor
, and the other is the amount of cur-
rent to charge the compensation capacitor
[10]–[15]. In
mathematical expression, the slew rate
is expressed as
(26)
From the above expression, the increase in the slew rate can be
achieved by either increasing the bias current or reducing the
value of the compensation capacitor. Increasing the bias current
will increase the static power consumption, so it is not preferred
in low-power design. Therefore, the most efficient method to
increase the slew rate is to reduce the value of the compensa-
tion capacitor. However, the compensation capacitor cannot be
reduced arbitrarily; otherwise, the amplifier will be unstable in
the close-loop operation.
By comparing (22) with (2), as the required value of the com-
pensation capacitor in DFCFC is reduced by
times, the im-
provement on the slew rate using DFCFC is improved by
times. Similar to the frequency response, the improvement by
DFCFC on the slew rate is larger for a large
to ratio.
Therefore, the slew rate of a DFCFC amplifier is better than that
of the NMC counterpart, especially when driving a large capac-
itive load.
E. Settling Time
Settling behavior of an amplifier is very essential in analog
design since it directly affects the performance of the circuits
using the amplifier. To enhance the settling behavior, the first
thing to investigate is its dependence. Neglecting the short
duration of the small-signal response before slewing, the
settling time
can be mainly divided into two periods.
They are the slewing period
and quasi-linear period
[10], [11]. The dependence of the slew rate has been
discussed in the previous section, and the slew rate is proven
to be greatly improved by DFCFC. For the quasi-linear period,
it is a strong function of the phase margin [17], and it also
depends on how compressed pole–zero doublet is below the
unity-gain frequency of the amplifier [1], [3].
In the case of using DFCFC, as there is no pole–zero doublet,
the quasi-linear period only depends on the phase margin. By
using the stability criteria stated before, the phase margin of the
DFCFC amplifier is approximately 60
, which is a very reason-
able value to obtain a short quasi-linear period. Asa result, there
are not many decaying sinusoidal cycles, and the output voltage
can settle within a very short duration.
F. Power Supply Rejection Ratio
The power supply rejection ratio
of a single-stage
cascode amplifier is excellent as the cascode configuration
provides a good isolation between the supply and the output of
the amplifier. However, the
in a noncascode multistage
amplifier is inferior due to the poorer isolation barrier of a
Fig. 5. High-frequency gain of the NMC and DFCFC amplifiers.
single transistor. Moreover, when the compensation network
is required, the compensation capacitors are short-circuited
at high frequencies, and the high-frequency supply variations
easily bypass through the compensation capacitor to the output
of the amplifier.
The
is defined by the following expression [15]:
(27)
where
is the open-loop input-to-output voltage gain,
and
is the supply-to-output voltage gain. It
can be shown from the above equation that the
can
be improved by obtaining a larger
and a smaller
. A larger can be achievedby increasing
the gain-bandwidth product of the amplifier. As illustrated in
Fig. 5, at a particular frequency
, the gain obtained from
the DFCFC amplifier is
, which is much
larger than
of the NMC amplifier. Therefore,
compared with NMC, the signal from the input of the DFCFC
amplifier is amplified larger, and the output signal appearing
at the output is stronger. As a result, the relative effect of the
supply noise is less significant.
A reduction in
can be accomplished by using
smaller values of compensation capacitors or completely elimi-
nating the capacitors. In the DFCFC topology, as no inner com-
pensation capacitor is required, the resulting
is
much smaller.
Taking into account both the above-mentioned advantages,
the
given by DFCFC is better than that of NMC.
G. Physical Dimension of the DFCFC Amplifier
In the design of multistage amplifiers that drive large capaci-
tiveloads, most ofthechip area isoccupiedby the compensation
capacitors. As the proposed topology requires smaller compen-
sation capacitors to achieve stability, the required dimension of
the whole amplifier is significantly reduced.
IV. I
MPLEMENTATION OF THE PROPOSED STRUCTURE
Low-power 2-V DFCFC and NMC amplifiers, illustrated
in Fig. 6, driving two capacitive loads, 100 and 1000 pF,
were fabricated using a double-metal double-poly 0.8-
m
CMOS process with
V and V from
LEUNG et al.: THREE-STAGE LARGE CAPACITIVE LOAD AMPLIFIER WITH DFCFC 227
(a)
(b)
Fig. 6. Circuit diagram of the (a) NMC amplifier and (b) DFCFC amplifier.
Fig. 7. Micrograph of the amplifiers.
AMS.
1
The micrograph is shown in Fig. 7. For the NMC
amplifiers, the first, second and third stages are implemented
by M101-M108, M201-M204, and M301, respectively. For
the DFCFC counterparts, the additional M302 is the FTS,
and M401-M404 is the DFC block. It should be noted that
since the circuits are to demonstrate the proposed frequency
compensation structure, the circuit structure displayed in Fig. 6
is one of many methods to implement the proposed structure.
Additional circuitry to control the dc operating point of the
DFC block and to implement a class-AB output stage with
feedback control can be added to the circuit structure to build a
higher performance amplifier.
To optimize the values of the compensation capacitors in the
DFCFC amplifiers, the values of
are 3 pF for both 100 and
1000 pF loading conditions. The values of
in both cases
are obtained from the previous analysis and are fine-tuned to
optimize the tradeoff between the bandwidth and phase margin.
As shown in Table I, since the values of the compensation ca-
pacitors are much smaller for DFCFC, the sizes of the DFCFC
1
Austria Mikro Systeme International AG, A-8141 Unterpremstatten, Aus-
tria.
228 IEEE TRANSACTIONS ON SOLID-STATE CIRCUITS, VOL. 35, NO. 2, FEBRUARY 2000
(a)
(b)
Fig. 8. Frequency responses of the (a) NMC and (b) DFCFC amplifier with
100-pF loading capacitance (only the frequencies near the unity-gain frequency
are shown).
Fig. 9. Transient responses of the amplifiers driving 100 pF.
amplifiersare small and easy to integrate. On the other hand, the
NMC amplifier drivinga 1000-pFload requiresunrealistic large
compensation capacitors
pF and pF),
and thus integration is impossible.
V. E
XPERIMENTAL RESULTS
The frequency responses measured with the HP4194A
impedance/gain-phase analyzer are shown in Figs. 8 and 10,
while the transient responses measured with the LeCroy 9354A
oscilloscope are shown in Figs. 9 and 11. The performances are
tabulated in Table I.
(a)
(b)
Fig. 10. Frequency responses of the (a) NMC and (b) DFCFC amplifier
with 1000-pF loading capacitance (only the frequencies near the unity-gain
frequency are shown).
Fig. 11. Transient responses of the amplifiers driving 1000 pF.
As shown in the figures and table, by comparing the DFCFC
amplifier to the NMC counterpart, DFCFC improves the gain-
bandwidth product by 4 times, the slew rate by 6 times, the set-
tling time by 3 times, and the negative power supply rejection
ratio by at least 40 dB for the case driving 100 pF. For a larger
loadingcapacitance of 1000pF, the improvement is moresignif-
icant. The gain-bandwidthproduct isimprovedby 18 times. The
slew rate and settling time are improved by 14 and 9 times, re-
spectively. In addition, the negativepower supply rejection ratio
is enhanced by at least 60 dB. However, there is only negligible
increase on the power consumption.
LEUNG et al.: THREE-STAGE LARGE CAPACITIVE LOAD AMPLIFIER WITH DFCFC 229
TABLE I
M
EASURED RESULTS OF THE NMC AND DFCFC AMPLIFIERS
TABLE II
C
OMPARISON OF DIFFERENT MULTISTAGE AMPLIFIERS
Sincethe bandwidthimprovementby DFCFCwith a 1000-pF
load is 18 times, which is much greater than 4, the bandwidth
provided by the amplifier is even wider than a single-stage am-
plifier under nearly the same power consumption.
To provide a clearer picture on the improvements by DFCFC,
acomparisontableforsome published amplifiersusingdifferent
compensation topologies is shown in Table II. Two figures of
merit,
[6] and , are defined for small-signal and
large-signal performances.
(28)
(29)
The units of
and are MHz pF/mW and V/ s
pF/mW, respectively. An average is used in the calculation.
Alargerfigureof merit implies a betterfrequencycompensation
topology. From Table II, it is clear that DFCFC is better than all
existing topologies, especially for the case of 1000-pF loading
capacitance.
VI. C
ONCLUSION
In this paper, DFCFC which is targeted for amplifiers driving
large capacitive loads has been presented. Theoretical analysis
on frequencyresponse, transient response, and powersupply re-
jection ratio have been firstly discussed. Then, the experimental
results have demonstrated that a DFCFC amplifier is signifi-
cantly better than an NMC amplifier and a single-stage am-
plifier when driving large capacitive loads. Finally, comparison
withotherpublished topologies hasbeenpresented,and DFCFC
shows better small-signal and large-signal performances.
230 IEEE TRANSACTIONS ON SOLID-STATE CIRCUITS, VOL. 35, NO. 2, FEBRUARY 2000
ACKNOWLEDGMENT
The authors would like to thank Prof. W. T. Ng from the Uni-
versity of Toronto, Toronto, ON, Canada, for his valuable sug-
gestions, and theyalso would like to thank S. F. Luk and J. Chan
from HKUST for their technical assistance.
R
EFERENCES
[1] R. G. H. Eschauzier and J. H. Huijsing, Frequency Compensation Tech-
niques for Low-Power Operational Amplifiers. Boston, MA: Kluwer,
1995.
[2] J. H. Huijsing, R. Hogervorst, and K.-J. de Landen, “Low-power low-
voltage VLSI operational amplifier cells,” IEEE Trans. Circuits Syst. I,
vol. 42, pp. 841–852, Nov. 1995.
[3] R. G. H. Eschauzier, L. P. T. Kerklaan, and J. H. Huijsing, “A 100-MHz
100-dB operational amplifier with multipath nested Miller compensa-
tion structure,” IEEE J. Solid-State Circuits, vol. 27, pp. 1709–1717,
Dec. 1992.
[4] R. G. H. Eschauzier, R. Hogervorst, and J. H. Huijsing, “A pro-
grammable 1.5 V CMOS class-AB operational amplifier with hybrid
nested Miller compensation for 120 dB gain and 6 MHz UGF,” IEEE J.
Solid-State Circuits, vol. 29, pp. 1497–1504, Dec. 1994.
[5] F.You,S.H.K.Embabi,andE.Sánchez-Sinencio,“Multistageamplifier
topologies with nested G
-C compensation,” J. Solid-State Circuits,
vol. 32, pp. 2000–2011, Dec. 1997.
[6] H. T. Ng, R. M. Ziazadeh, and D. J. Allstot, “A multistage amplifier
techniquewith embedded frequency compensation,”IEEE J.Solid-State
Circuits, vol. 34, pp. 339–347, Mar. 1999.
[7] R. J. Widlar, “Design techniques for monolithic operational amplifiers,”
IEEE J. Solid-State Circuits, vol. SSC-4, pp. 184–191, Aug. 1969.
[8] K. N. Leung, P. K. T. Mok, W. H. Ki, and J. K. O. Sin, “Damping-
factor-control frequency compensation technique for low-voltage low-
power large capacitive load applications,” in ISSCC’99 Dig. Tech. Pa-
pers, 1999, pp. 158–159.
[9] G. C. Temes and J. W. LaPatra, Introduction to Circuit Synthesis and
Design, 1st ed. New York: McGraw-Hill, 1977.
[10] B. Y. Kamath, R. G. Meyer, and P. R. Gray, “Relationship between fre-
quency response and settling time of operational amplifier,” IEEE J.
Solid-State Circuits, vol. SCS-9, pp. 347–352, Dec. 1974.
[11] C. T. Chuang, “Analysis of the settling behavior of anoperational ampli-
fier,” IEEE J. Solid-State Circuits, vol. SSC-17, pp. 74–80, Feb. 1982.
[12] R. Klinke, B. J. Hosticka, and H.-J. Pfleiderer, “A very-high-slew-rate
CMOS operational amplifier,” IEEE J. Solid-State Circuits, vol. 24, pp.
744–746, June 1989.
[13] B. W. Lee and B. J. Sheu, “A high slew-rate CMOS amplifier for analog
signal processing,” IEEE J. Solid-State Circuits, vol. 25, pp. 885–889,
June 1990.
[14] P. R. Gray and R. G. Meyer, Analysis and Design of Analog Integrated
Circuits. New York: Wiley, 1984.
[15] R. Gregorian and G. C. Temes, Analog MOS Integrated Circuits for
Signal Processing, 1st ed. New York: Wiley, 1986.
[16] J. W. Nilsson, Electric Circuits. Reading, MA: Addison-Wesley, 1993.
[17] H.C. Yang and D. J.Allstot, “Considerations forfastsettling operational
amplifiers,”IEEE Trans. Circuits Syst., vol. 37, pp.326–334, Mar. 1990.
Ka Nang Leung received the B.Eng. and M.Phil.
degrees from the Department of Electrical and
Electronic Engineering at the Hong Kong University
of Science and Technology, Clear Water Bay,
Hong Kong, in 1996 and 1998, respectively. He is
now working toward the Ph.D. degree in the same
department.
During the B.S. studies, he joined Motorola,
Hong Kong, to develop a PDA system as his final
year project. His research interests include the
development of novel frequency compensation
techniques for low-voltage low-power multistage amplifiers, design of
low-dropout regulators, and implementation of PWM dc–dc converters for
telecommunication systems.
In 1996, he receivedaBest Teaching Assistant Award from the Department of
Electrical and Electronic Engineering at the Hong Kong University of Science
and Technology.
PhilipK. T. Mok(S’86–M’95) received the B.A.Sc.,
M.A.Sc.andPh.D.degreesinelectricalandcomputer
engineering from the University of Toronto, Toronto,
ON, Canada, in 1986, 1989, and 1995, respectively.
While at the University of Toronto, he was a
Teaching Assistantin both theElectrical Engineering
and Industrial Engineering Departments from 1986
to 1992. He taught courses in circuit theory, IC
engineering,andengineeringeconomics.Hewasalso
employed as a Research Assistant in the Integrated
Circuit Laboratory at the University of Toronto from
1992to1994.HejoinedtheDepartmentofElectricalandElectronicEngineering,
the Hong Kong University of Science and Technology, Hong Kong, in January
1995as anAssistantProfessor,whereheisnowalso theAssociateDirectorof the
Computer Engineering Program. His research interests include semiconductor
devices, processing technologies, and circuit designs for power electronics and
telecommunications applications, with current emphasis on power integrated
circuitsandlow-voltageanalogueintegratedcircuitsdesign.
Dr. Mok received the Henry G. Acres Medal, the W.S. Wilson Medal and
Teaching Assistant Award from the University of Toronto, and the Teaching
Excellence Appreciation Award from the HongKong University of Science and
Technology.
Wing-Hung Ki (S’86–M’92) received the B.Sc. de-
greefromtheUniversityof California, San Diego,the
M.Sc. degree from the California Institute of Tech-
nology (Caltech), Pasadena, the Engineer degree and
the Ph.D. degree from the University of California,
Los Angeles, in 1984, 1985, 1990, and 1995, respec-
tively, all in electrical engineering.
From 1985 to 1986, he was a Research Assistant
in the Power Electronics Group atCaltech, helping to
develop the software SCAMP (Switching Converter
Analysis and Measurement Program). In 1992, he
joined Micro Linear Corporation, San Jose, CA, as a Senior Design Engineer
in the Department of Power and Battery Management, working on the design
of power converter controllers. In 1995, he joined the Hong Kong University of
Science and Technology, Clear Water Bay, Hong Kong. His research interests
are design and modeling of dimmable electronic ballasts, switch mode and
switched-capacitor powerconverters, IC controller design forpowerconverters,
and fundamental research in analog integrated circuits.
Dr. Ki is the recipient of the 1997–1998 Asia Innovator Award of the Year
granted by EDN Asia.
Johnny K. O. Sin (S’79–M’88–SM’96) was born in
Hong Kong. He received the B.A.Sc. degree in 1981,
the M.A.Sc. degree in 1983, and the Ph.D. degree in
1988, all in electrical engineering, from the Univer-
sity of Toronto, Toronto, ON, Camada.
Upon completion of the Ph.D. degree, he joined
Philips Laboratories, New York, where he was a
Senior Member of the Research Staff from 1988 to
1991. He joined the Department of Electrical and
Electronic Engineering, the Hong Kong University
of Science and Technology, Clear Water Bay, Hong
Kong, in August 1991, as an assistant professor and was promoted to Associate
Professor in 1996. He is one of the founding members of the Department and
has been serving as the Director of the Undergraduate Studies program since
the fall of 1998. He holds three U.S. patents and has four patents pending. He
has published over 130 papers in technical journals and refereed conferences in
the above areas. His research interests lie in the general area of microelectronic
devicesandfabricationtechnologyandiscurrentlyworkingintheareasofpower
semiconductor devices and IC’s, thin-film transistors, field emission devices,
integratedsensors,Silicon-On-InsulatorRF,andpowerdevicesandtechnology.
Dr. Sin is a member of the EDS Power Devices and IC’s Technical Com-
mittee. He served as Technical Committee Member of the International Con-
ference on Microelectronics Test Structures (ICMTS). He is also a Technical
Committee Member of the International Symposium on Power Semiconductor
Devices and IC’s (ISPSD). Dr. Sin was made an Honorary Visiting Professor of
the Dalian University of Technology, Dalian, R.O.C., in 1996. He is an Editor
forIEEEE
LECTRONDEVICE LETTERS.He was awardedtheTeaching Excellence
Appreciation Award by the School of Engineering, the Hong Kong University
of Science and Technology, in 1998.
