You are on page 1of 8

Microelectronics Journal 42 (2011) 758765

Contents lists available at ScienceDirect

Microelectronics Journal
journal homepage: www.elsevier.com/locate/mejo

A novel architecture for improving slew rate in FinFET-based op-amps


and OTAs
Rajesh A. Thakker a,, Mayank Srivastava b, Ketankumar H. Tailor a, Maryam Shojaei Baghini a,
Dinesh K. Sharma a, V. Ramgopal Rao a, Mahesh B. Patil a
a
Department of Electrical Engineering, Indian Institute of Technology (IIT), Bombay, India
b
Senior Engineer, 28nm/20nm ESD Development, Intel Mobile Communications, USA

a r t i c l e i n f o a b s t r a c t

Article history: A new architecture for improvement of slew rate (SR) of an op-amp or an operational transconductance
Received 23 September 2010 amplier (OTA) in FinFET technology is proposed. The principle of operation of the proposed
Received in revised form architecture is based on a set of additional current sources which are switched on, only when OTA
12 January 2011
should provide a high current, usually for charge or discharge of large load capacitor. Therefore, the
Accepted 17 January 2011
Available online 17 February 2011
power overhead is less compared to conventional high SR designs. The commonly used two-stage
Miller-compensated op-amp, designed and optimized in sub 45 nm FinFET technology with 1 V single
Keywords: supply voltage, is used as an example for demonstration of the proposed method. For the same FinFET
High slew rate op-amp technology and with optimal design, it is shown that the slew rate of the op-amp is signicantly
Look-up table
improved. The slew rate is improved from 273 to 5590 V=ms for an input signal with a rise time of
FinFET
100 ps. The other performance measures such as gain and phase margin remain unchanged with the
Hierarchical particle swarm optimization
Circuit design additional circuitry used for slew rate enhancement.
& 2011 Elsevier Ltd. All rights reserved.

1. Introduction Its slew rate degrades as output capacitance charges (discharges)


or as supply voltage reduces. In [4], an OTA cell with high slew
In applications like switched-capacitor lters or switched op- rate was implemented in 0:5 mm CMOS technology. It used a class
amps, fast switching of output voltage from an initial value to a AB input stage to control the slew rate. However, the amount of
nal value is desired. Therefore, settling time of the output should slew rate enhancement depends on the value of input differential
not be limited by the slew rate of the op-amp. This motivates voltage. It could achieve a maximum slew rate of 100 V=ms at
developing special circuit structures which improve op-amp 80 pF load with 0.12 mW power dissipation. The class A/AB and
slew rates. AB/AB two-stage op-amps was fabricated in 0:5 mm CMOS tech-
Table 1 shows comparison of important performance para- nology which achieved a symmetrical slew rate of 16 V=ms [5]. A
meters of few of the reported slew rate-enhanced circuits in fast-transient switching DCDC converter using a DS modulator,
different CMOS technologies. A CMOS op-amp, implemented in designed in 0:35 mm CMOS technology, could achieve 0:56 V=ms
6 mm CMOS technology with internal transistor compensation, slew rate at 10 nF load capacitance with a power dissipation of
and slew rate of 36=50 V=ms with power dissipation of 200 mW [6]. In [7], a low power class-AB CMOS OTA with rising
11.5 mW has been reported in early 80s [1]. In [2], a class AB and falling slew rates of 4.92 and 5:04 V=ms, respectively, at 10 pF
op-amp using common mode feedback, implemented in 2 mm load capacitance was designed and implemented in 0:18 mm
technology, has been demonstrated. This op-amp exhibits CMOS technology. In [8], a fast settling slew rate enhancement
2:5 V=ms slew rate at 10 pF load capacitance while dissipating technique for an op-amp using constant-gm biasing was designed
0.1 mW power. In [3], an input differentiator is used to turn on an in 0:18 mm process. It used an auxiliary op-amp which slews at
additional bias current source at the input stage whenever the the same time as the main op-amp slews. It increased slew rate
rate of change of input voltage is larger than preset value. The from 25 to 150 V=ms with a total static power dissipation of
circuit proposed in [3] was implemented in 0:8 mm CMOS tech- 5.8 mW. A SR controlled output driver which uses a phase-locked
nology and could achieve a positive slew rate of 2667 V=ms at a loop was reported in [9] and implemented in 0:18 mm CMOS
load capacitance of 15 pF at the cost of 14 mW power dissipation. technology. It could achieve a slew rate of 0.41 V/ns for a load
capacitance of 1540 pF. In [10], a two-stage op-amp with both
gain and slew rate enhancement was simulated in 0:18 mm CMOS
 Corresponding author. Tel.: + 91 9426333130. technology. It could achieve 26:8 V=ms slew rate and 74 dB DC
E-mail address: rathakker2008@gmail.com (R.A. Thakker). gain with 362 mW power consumption. A 1 V digital OTA for

0026-2692/$ - see front matter & 2011 Elsevier Ltd. All rights reserved.
doi:10.1016/j.mejo.2011.01.010
R.A. Thakker et al. / Microelectronics Journal 42 (2011) 758765 759

Table 1
Comparison of reported work of slew rate enhanced circuits with that demonstrated in this paper. Here, Av voltage gain, SR slew rate, CL load capacitance, PD power
dissipation, ts settling time, UGB unity gain bandwidth, PM phase margin, E/S experimental/simulation.

Technology Av SR, CL PD ts UGB PM E/S

6 mm [1] Z 60 dB 36=50 V=ms, 32 pF 11.5 mW 0:25 ms 8 MHz 501 E


2 mm [2] 70 dB 2:5 V=ms, 10 pF 0.1 mW E
0:8 mm [3] 100.7 dB Z 2 V=ns, 15 pF 14.1 mW + 15 ns/  19 ns 55 MHz S
0:5 mm [4] 43 dB 36=50 V=ms, 80 pF 120 mW + 29 ns/  57 ns 725 kHz 89.51 E
0:5 mm [5] 16 V=ms, 10 pF 15 MHz E
0:35 mm[6] 0:56 V=ms, 10 nF 200 mW 200 MHz 501 S
0:18 mm [7] 48.97 dB 4:92=5:04 V=ms, 10 pF 1:96 mW 2:1 ms 57.27 kHz 78.181 S
0:18 mm [8] 150 V=ms,- 5.76 mW S
0:18 mm [9] 1 V/ns, 40 pF E
0:18 mm [10] 74 dB 26:8 V=ms, 1.75 pF 362 mW 160 MHz S
0:13 mm [11] 63.5 dB 16:29 V=ms, 34 pF 82 mW 734 ns 19.2 MHz 74.951 S
90 nm [12] 17 dB 1354 V=ms, 11.5 GHz 61.21 E
45 nm, this work 83 dB 1043=176 V=ms, 1 pF 103 mW + 3.98/ 19.2 ns 77.4 MHz 601 S
79.65 dB 6171=5590 V=ms, 20 fF 78 mW + 310/  600 ps 695 MHz 611 S

switched capacitor applications with a slew rate of 16:29 V=ms that the proposed technique also works for fully differential
and power dissipation of 82 mW was realized in 0:13 mm CMOS FinFET OTAs with common mode feedback (CMFB) without effect
technology [11]. In [12], two-stage Miller and folded cascode op- on the stability of the OTA. The proposed architecture can be used
amps were programmed digitally to enhance the gain, slew rate for planar MOSFET as well. It is well known that multigate devices
and compensation, and were implemented in standard 90 nm such as FinFETs have the advantage of very less short-channel
CMOS process. The performance parameters of folded cascode op- effects compared to planar MOSFETs. Multigate devices are one
amp [12] with programmable characteristics are mentioned category of promising devices in near future and therefore, in this
in Table 1. paper, we have tested the proposed architecture using FinFET
There is some reported literature for slew rate improvement of devices.
BJT circuits. A high slew rate op-amp with class AB input stage The paper is organized as follows. In Section 2, the basic
was fabricated in 40 V complementary bipolar technology [14]. A principle of the proposed architecture is described. The circuit
symmetrical BJT voltage follower with high slew rate of schematic and simulation results for the two-stage op-amp,
4950 V=ms was reported in [15]. with and without slew rate enhancement circuitry, are given
There are also reported patents which focus on the class AB in Section 3. The simulation results for the differential amplier
operation of input circuit to improve SR of the op-amp or OTA. with CMFB circuit in combination with the slew rate enhance-
Class AB operation reduces power overhead compared to the ment are presented in Section 4. Finally, conclusions are sum-
conventional high-bias current design. In [16], improvement in marized in Section 5.
sinking of load current and ability to drive large capacitive loads
were achieved by inserting source follower stage between folded
cascoded gain stage and class AB stage. A phase inverter circuit 2. Basic principle
was used to control the current source circuits to obtain a high
slew rate over the voltage range of approximately 4 V [17]. A The slew rate of an OTA or op-amp is proportional to the
controllable assistant output stage was used in addition to the maximum current, usually available from the rst stage of the
output stage in [18] to improve the slew rate. The output buffer circuit. Increase in the slew rate requires increase in the value of
and feedback circuitry-based slew rate control circuit, which also bias current source, which will increase the overall power
increases gate-oxide reliability of the IC is described in [19]. dissipation of the circuit. Besides other performance measures
This paper presents a new power-efcient architecture for of the op-amp will get affected. The objective of the basic idea,
improvement of slew rate (SR) of an op-amp or an OTA. It is tested presented in this paper, is to achieve a high slew rate, with a low
in 45 nm FinFET technology at a supply voltage of 1 V. Previously power overhead while maintaining other performance measures
reported papers on slew rate improvement have been demon- of op-amp (or OTA) unchanged. The principle of operation is
strated in CMOS technologies older than 45 nm [119] and most detecting the onset of slewing operation of the circuit and
of them are reported to operate at a supply voltage higher than accordingly activating the additional current sources in the
1 V. The circuit demonstrated in [11] works at 1 V but it offers appropriate directions. The additional current sources are acti-
very small slew rate of 16:29 V=ms. The digitally programmed vated only during the slewing operation of the op-amp circuit,
folded cascode op-amp, implemented in 90 nm CMOS technol- and hence power dissipation and other performance measures of
ogy [12], resulted in maximum slew rate of 1354 V=ms for DC the basic op-amp circuit mostly remain unchanged.
voltage gain of 17 dB. The proposed architecture in this paper The block diagram of the novel SR improvement circuit is
enhances the slew rate by a set of additional current sources shown in Fig. 1. There are mainly two sections in the proposed
which are controlled by comparators. The comparators turn architecture, which are used for slew rate improvement. These
current sources on if absolute value of differential input voltage sections constitute unsymmetrical comparators as detection cir-
exceeds some threshold value during charge or discharge of cuits and additional switchable current sources as current boost-
circuit capacitor(s). It is also shown that the inclusion of com- ing modules. As shown in the gure, additional current sources, I1
parators and current sources does not signicantly affect other to I4, are controlled by comparators: COMP1 and COMP2. These
performance measures except power dissipation of OTA/op-amp. comparators are sized in such a way to keep the output of the
The power dissipation of op-amp/OTA also increases marginally control logic circuit at a proper logic level when the absolute
compared to the relative increase in the slew rate. We also show value of the differential input voltage is less than some threshold
760 R.A. Thakker et al. / Microelectronics Journal 42 (2011) 758765

value (denoted by Vsw). I1 and I3 are controlled by COMP1 while I2 the output of COMP2 becomes high, which activates I2 and I4
and I4 are controlled by COMP2. I1 and I3 current sources are current sources and accordingly negative slew rate improves.
approximately equal in terms of value, and similarly, I2 and I4
current sources. During positive slewing operation, i.e., when Vin1 is
greater than Vin2 by some marginal value, comparator COMP1 3. Circuit schematic and simulation results
operates, and in negative slewing operation, i.e., when Vin1 is less
than Vin2 by some marginal value, comparator COMP2 operates. 3.1. FinFET technology
The output of COMP1, C 1 , goes low when the difference between
Vin1 and Vin2 exceeds certain voltage (i.e., switching voltage, Vsw) The principle explained in Section 2 and shown in Fig. 1, is
and hence current sources I1 and I3 are activated. The activation of implemented in 45 nm technology FinFET devices. The TCAD
I1 and I3 current sources, injects additional current into the simulator [20] mobility parameters are rst tuned to match with
compensation capacitor Cc. The activation of two current sources FinFET device experimental data (Fig. 2) [21,22]. The FinFET tech-
together completes the path from supply to Vss for charging Cc and nology parameters used in this work are: minimum channel length
therefore, speeds up positive slewing operation. The value of Vsw is (L) is 20 nm, effective oxide thickness (EOT) is 1.6 nm, n width
a design variable, which in fact determines the switching point of (WFIN) is 6 nm, and n height (HFIN) is 30 nm. The channel doping is
the detector circuit. Similarly, during negative slewing operation, 1  1015 cm  3 and the source/drain doping is 1  1020 cm  3 with

Additional Current
Sources Comparators
Twostage Opamp VDD COMP1

R Cc I1 Vin1

C1 Vin2

Vin1

Vin2 Vout COMP2

CL I2 Vin2
C2
A
I3 I4 Vin1
C1 C2

Fig. 1. Architecture of the SR improvement module employed in a two-stage op-amp circuit.

Fig. 2. Calibration of TCAD models for drift diffusion transport with experimental data [21,22].
R.A. Thakker et al. / Microelectronics Journal 42 (2011) 758765 761

an overlap distance (LOV) of 2 nm and a 1 nm/decade Gaussian op-amp, shown in Fig. 3(a), is designed and optimized for maximizing
doping gradient into the channel. slew rate under a given power constraint (o 50 mW). The power
dissipation constraint considered in the design is less than some of
the earlier reported work [14,6,10] on slew rate improvement. The
3.2. Circuit schematic and design circuit schematics for comparators (COMP1 and COMP2) along with
the switchable current sources are given in Fig. 3(b) and (c). Look-up
The circuit schematic, used to verify the proposed principle, is table (LUT) approach [23] implemented in the SEQUEL circuit
shown in Fig. 3. As an example, the commonly used two-stage simulation package [24] is used for simulation of the circuits in this

V
DD
(18) (18) (162)

M M Cc M
3 4 5
V out R
1
V in2 (7.25 fF)
C
V in1 L
(20 fF)
M M
1 (11) A (11) 2
M M M 8 (20)
6 (2) 7 (9)

V V V
DD DD DD
M M M M M M
12 13 14 15 17 (1) 18 (1)
V
DD

(1) (1) (1) (1) C1 C1


I
1
M
20 (2)
V in2
V in1 V out
M (1) M (1)
M (15) (15) M 11 16 19
10 A
M
9 (12)
I
B 3
M
21 (2)

V V
DD DD
M M M M M
25 26 27 28 30 (1)
V out
M
(1) (1) (1) (1) 32 (3)
C2
I
2

V in1 V in2
M (1) A
M (15) (15) M 29
23 24 M
31 (3)
M
22 (12) I
4
B

Fig. 3. The circuit schematic for slew rate improvement using FinFETs: (a) two-stage op-amp circuit, (b) unsymmetrical comparator COMP1 and I1 and I3 current sources
and (c) unsymmetrical comparator COMP2 and I2 and I4 current sources. Numbers in brackets show number of ns for FinFET devices, and values for resistors and
capacitors. Length of input transistors of two comparators is given in (b) and (c). Channel length of other transistors is 50 nm.
762 R.A. Thakker et al. / Microelectronics Journal 42 (2011) 758765

work. A hierarchical particle-swarm-optimization algorithm is used from 46 to 78 mW. The increase in power dissipation is marginal in
for optimization of FinFET circuits for desired specications as comparison to the improvement observed in slew rates. Other
described in detail in [23]. specications of op-amp are not affected. For obtaining DC power
Op-amp circuit is designed using FinFET devices of channel dissipation (no signal) both inputs, Vin1 and Vin2 are connected to
length 50 nm. To obtain the required value of switching (sensing) VDD/2. For obtaining average power dissipation, a square wave with
voltage (Vsw), it is desirable to use asymmetric comparators, for frequency of 100 MHz, and rise and fall time of 100 ps has been
example, by using asymmetric input devices. Asymmetry in the applied to the input. Obtained values are: 46 mW without slew rate
comparator may be implemented in different ways, e.g., by enhancement circuitry and 71:7 mW with slew rate enhancement
introducing mismatch between lengths or number of ngers of circuitry.
two corresponding input transistors. In this work, FinFET devices To nd out, the best possible slew rate which can be obtained
with different channel lengths are used in the input stage to with the conventional two-stage op-amp without using any
achieve the required asymmetry. This means input transistors additional circuitry for slew rate enhancement, the two-stage
M10 and M11 have different channel lengths in COMP1 and op-amp was designed without any constraint on power dissipa-
similarly, transistors M23 and M24 in COMP2 are of different tion. It gave a slew rate of 650 V=mS with a power dissipation of
channel lengths. We have chosen 50 nm channel length for M10 113 mW. This circuit is referred as two-stage opamp-2. This
and remaining transistors in COMP1 except M11 transistor. FinFET power dissipation is higher than the proposed circuit which is a
device with 100 nm channel length is chosen for M11 transistor. combination of slew rate enhancement circuitry and two-stage
Similarly, in COMP2, 50 nm channel length for M23 and remaining opamp-1. The slew rate for two-stage opamp-2 is also almost
transistors in COMP2 except M24. The channel length of transis- one-tenth of the proposed circuit.
tors M24 and M11 is 100 nm. Using this approach, COMP1 and The transient response for all three circuits, two-stage opamp-1,
COMP2 are made unsymmetrical. Look-up tables are generated two-stage opamp-2, and the proposed circuit (SR-improved op-
using TCAD simulator for 50 and 100 nm devices to provide look- amp), is shown in Fig. 4 for an input signal with a rise time of 100 ps.
up tables for simulation of circuits using LUT simulator. It can be seen that the proposed circuit shows a very high slew rate
The automatic design approach (LUT-based optimizer), pro- in comparison to other two circuits.
posed by the authors for the design and optimization of FinFET- The slew rates of two-stage opamp-1 and proposed circuit are
based circuits, is used to efciently design and optimize all circuits measured for an input signal with rise and fall time varying in the
demonstrated in this work [25]. Initially, the op-amp circuit is range of 25 ps to 2 ns. The simulation results are shown in Fig. 5.
automatically designed using LUT-based optimizer for specica-
tions given in Table 2. The resulted design (number of ns for
FinFETs and other component values) is given in Fig. 3(a) and 0.8
achieved specications are shown in Table 2. The slew rate values, Input
shown in Table 2, are obtained by keeping op-amp in unity gain
0.7
conguration and applying input pulse signal with a rise and fall
time of 100 ps. This circuit is referred as two-stage opamp-1.
Volts (V)

0.6 Output: Proposed


By running multiple design and optimization trials for two-
Circuit
stage opamp-1, it was veried that the obtained design speci-
Output: Twostage
cations, given in Table 2, are the best possible values. Comparator 0.5 Opamp1
circuit is designed for the switching voltage Vsw 55 mV, using
Output: Twostage
automatic design approach. It should be noted that input transis- 0.4 Opamp2
tors of comparators have different channel lengths. The design for
unsymmetrical comparators, i.e., number of ns for FinFET
0.3
devices, is given in Fig. 3(b) and (c). The number of ns for FinFET
devices in current sources is chosen to obtain best settling time 0 1 2 3 4 5 6 7 8
during slewing operation. Time (ns)
The simulation results of op-amp circuit in combination with
Fig. 4. The transient response of two-stage opamp-1, two-stage opamp-2, and the
slew rate enhancement circuitry, referred as SR-improved op-amp, proposed circuit for an input signal with a rise time of 100 ps.
are given in Table 2. It can be seen from the table that slew rate is
improved from 271 to 5590 V=mS and power dissipation increases

104
Slew Rate: Proposed Circuit
Slew Rate: Twostage
Table 2 Opamp1
Desired and obtained specications of two-stage op-amp circuit of Fig. 3 without
and with slew rate enhancement circuitry. Op-amp circuit without slew rate
enhancement circuitry (i.e., Fig. 3(a)) is referred as Two-stage Opamp-1. The
circuit with slew rate enhancement is referred as SR-improved op-amp. All 103
circuits are automatically designed and optimized.

Specications and desired values Best obtained values SR-improved


(two-stage opamp-1) op-amp

Gain Z 80 dB 79.65 dB 79.65 dB


Phase margin Z 651 611 611
Unity gain frequency Z 500 MHz 695 MHz 695 MHz 102
10 100 1000 10000
Power dissipation r 50 mW 46 mW 78 mW
Offset voltage r 50 mV 83 mV 83 mV Input signal rise time (ps)
Slew rate rise Z 500 V=mS 273 V=mS 6171 V=mS
Slew rate fall Z 500 V=mS 271 V=mS 5590 V=mS Fig. 5. The comparison of slew rates of two-stage opamp-1 and proposed circuit
for an input signal with rise and fall time varying in the range of 25 ps to 2 ns.
R.A. Thakker et al. / Microelectronics Journal 42 (2011) 758765 763

It can be seen that the slew rate of two-stage opamp-1 is adjustment of number of ns in FinFET devices used as current
independent of input rise time. However, the slew rate of the sources. Therefore, the appropriate settling time is achieved.
proposed circuit, increases as rise time of the input signal Delay of comparators is another controlling factor which trades
decreases down to 100 ps and after that, it almost remains with the achieved slew rate.
constant with marginal increase. The proposed circuit architec- We also tested the proposed architecture for a load capaci-
ture is able to follow input signal until rise and fall time of input tance of 1 pF. The two-stage op-amp was redesigned for load
signal is decreased to 100 ps. Once it is reduced below 100 ps the capacitance of 1 pF, and the comparator circuit shown in Fig. 3
comparator fails to detect difference of input and output signals was ne tuned for this new load capacitance. It increases the slew
at the same rate at which input signal changes. Therefore, rate from 27.48 to 1043 V=ms for rising edge of input signal and
comparator is not able to switch ON and switch OFF additional from 22.19 to 176 V=ms for falling edge of input signal. The power
current sources at the same transition rate at which input signal dissipation was increased from 57 to 103 mW. Other specica-
changes (below 100 ps). Hence no more improvement is observed tions such as phase margin 601, unity gain frequency
in the slew rate of the circuit. (77.4 MHz), gain (83 dB), etc., were unchanged.
Table 3 shows the settling time of all three circuits, SR-
improved Op-amp, Two-stage Opamp-1, and Two-stage
Opamp-2. Settling time is measured as the time required for
the output to reach and remain within 2% of its nal value. It is 4. Differential amplier with CMFB circuit
measured for a load capacitance equal to 20 fF. It can be seen from
the table that the settling time of SR-improved Op-amp is In this section, we demonstrate the application of the slew rate
very small than that of Two-stage Opamp-1. The settling time improvement architecture for a differential amplier with com-
of SR-improved Op-amp is also better than that of Two-stage mon mode feedback (CMFB) circuit. The results show that the
Opamp-2 on the rising edge of input signal but they are stability of the differential amplier with CMFB circuit does not
comparable on the falling edge of input signal. As seen earlier, get affected due to the inclusion of slew rate improvement
the Two-stage Opamp-2 consumes more power, i.e., 113 mW in architecture. The differential amplier with CMFB circuit reported
comparison to 78 mW of SR-improved Op-amp circuit. There in [26] and shown in Fig. 6 is considered as an example in this
are some overshoot and undershoot in the transient response work. This circuit is designed in 45 nm FinFET technology. The
shown in Fig. 4. Such transients occur because enhanced slew rate channel length for all FinFET devices is taken as 20 nm. The
of op-amp (or OTA) is quite high, and hence by the time the specications of the designed circuit with and without slew rate
corresponding comparator turns off, output goes beyond desired enhancement are given in Table 4, and the number of ns for
value. Overshoots and undershoots are controlled by adjusting FinFET devices is shown in brackets in Fig. 6.
value of current sources. In this work, we minimized it with The slew rate improvement circuit (Fig. 3(b) and (c)) is
redesigned for this example (circuit in Fig. 6). It can be seen
from Table 4 that the slew rate improves from 875 to 1830 V=mS
Table 3 for the rising edge, and 775 to 1120 V=mS for the falling edge of
Settling time in pico seconds (ps) for SR-improved Op-amp, Two-stage Opamp- the input signal with the incorporation of slew rate enhancement
1, and Two-stage Opamp-2 for a load capacitance of 20 fF and for an input circuitry. The improvement in slew rate comes at the cost of
signal with rise/fall time of 50 ps. Here, ts,r stands for settling time on rising edge increase in power dissipation from 100 to 175 mW without
of input signal and ts,f for settling time on falling edge of input signal.
changing gain. The improvement in slew rate is not as good as
SR-improved Two-stage Two-stage the improvement, observed for the two-stage op-amp circuit
op-amp opamp-1 opamp-2 (see Table 2), where the slew rate is increased from 273 to
5590 V=mS. This is because the differential amplier with CMFB
ts,r 310 1490 620
circuit includes three transistors, controlled by CMFB signal,
ts,f 600 1610 660
operating as current sources.

VDD

(1) VCMFB (1) M M (1)


8 10

M (2) M M (2)
1 3 4
VCMFB
20 fF
(20) (20) 20 fF (1) (1)

M M M M
6 9 11
5
M
2 (2) (4)
VCMFB

M
7

Fig. 6. Differential amplier with common mode feedback (CMFB) circuit. The numbers in bracket shows the number of ns for the FinFET device. The channel length of
FinFET device is 20 nm.
764 R.A. Thakker et al. / Microelectronics Journal 42 (2011) 758765

Table 4 5. Conclusions
Specication of differential amplier (Fig. 6) with and without the slew rate
improvement architecture. Settling time is measured for a step input signal with a
A novel architecture and technique for improving slew rate of
rise and fall time of 50 ps.
op-amps or OTAs is presented. The proposed architecture is
Specications Differential Differential amplier with realized for two sample circuits optimized in 45 nm FinFET
amplier slew rate enhancement technology: (i) two-stage op-amp and (ii) differential amplier
with CMFB circuit. The slew rate of two-stage op-amp is increased
Gain 29.22 dB 29.22 dB
Phase margin 901 901
from 273 to 5590 V=ms with a marginal increase in power
Unity gain frequency 5.1 GHz 5.1 GHz dissipation from 46 to 78 mW for an input signal with a rise time
CMRR 84.96 dB 84.96 dB of 100 ps. The slew rate of differential amplier with CMFB
Power dissipation 100 mW 175 mW improved from 875 to 1830 V=mS for the rising edge, and 775 to
Offset voltage 83 mV 83 mV
1120 V=mS for the falling edge of the input signal with the
Slew rate rise 875 V=mS 1830 V=mS
Slew rate fall 775 V=mS 1120 V=mS incorporation of slew rate enhancement circuitry in the second
Rise settling time 215 ps 75 ps circuit example. DC gain and stability of differential amplier
Fall settling time 395 ps 415 ps with CMFB circuit are not affected with the inclusion of slew rate
improvement architecture.

Diff. Amp. Output


Acknowledgments
Input
0.6
The authors would like to thank Michael Fulde from Inneon
for useful discussions and Center for Nanoelectronics, Department
Volts (V)

of Electrical Engineering, IIT-Bombay for support. Synopsis is


0.5 gratefully acknowledged for Sentaurus licenses. Patent applica-
tion for the proposed idea has been led under application no.
IN-800533 for Indian and US patents.

0.4 References
Diff. Amp. with slew rate
enhancement architecture
[1] B. Wu, J. Mavor, High slew rate CMOS operational amplier employing
0 1 2 internal transistor compensation, Microelectronics Journal 14 (5) (1983)
Time (ns) 513.
[2] M.V. Ivanov, M. Ismail, V.N. Ivanov, High slew rate micro-power CMOS OTA
Fig. 7. The gure shows the transient response of differential amplier with and with class AB input stage, in: Proceedings of the 40th Midwest Symposium on
without slew rate improvement architecture for an input signal with a rise and fall Circuits and systems, vol. 2, 1997, pp.11971200. doi:10.1109/MWSCAS.
time of 50 ps. 1997.662294.
[3] P.C. Subramaniam, et al., High slew-rate CMOS operational amplier, Elec-
tronics Letter 39 (8) (2003) 640641.
The settling time for the differential amplier with and with- [4] A.J. Lopez-Martin, S. Baswa, J. Ramirez-Angulo, R.G. Carvajal, Low-voltage
out slew rate improvement architecture is measured for an input super class AB CMOS OTA cells with very high slew rate and power efciency,
IEEE Journal of Solid-State Circuits 40 (5) (2005) 10671077.
signal with a rise and fall time of 50 ps. It can be seen [5] J. Ramirez-Angulo, R.G. Carvajal, A. Lopez-Martin, High slew rate two stage
from Table 4 that the settling time improves considerably for A/AB and AB/AB op-amps with phase lead compensation at output node and
the rising edge of the input signal, whereas it slightly deteriorates local common mode feedback, in: IEEE International Symposium on Circuits
and Systems (ISCAS), 2008, pp. 288291, doi:10.1109/ISCAS.2008.4541411.
for the falling edge of the input signal. The one reason is the same;
[6] M. Song, D. Ma, A fast-transient over-sampled delta-sigma adaptive DCDC
the unity gain frequency of the differential amplier is a very converter for power-efcient noise-sensitive devices, in: ACM/IEEE Interna-
high. The second reason is, the input common mode range of the tional Symposium on Low Power Electronics and Design (ISLPED), 2007, pp.
286291.
comparator is not sufcient. The second reason can be taken care
[7] A.-R. Kim, H.-R. Kim, Y.-S. Park, Y.-K. Choi, B.-S. Kong, Low-power class-AB
by using the comparators with better or rail-to-rail input com- CMOS OTA with high slew-rate, in: International Symposium on System on
mon mode range. It should be noted that the average of settling Chip (SoC), 2009, pp. 313316.
time at rising and falling edge is better in case when the slew rate [8] S. Gopalkrishna, G. Krishna, B. Jalali-Farahani, A fast settling slew rate
enhancement technique for operational ampliers, in: 53rd IEEE Interna-
improvement circuit is present. The transient response of the tional Midwest Symposium on Circuits and Systems, 2010, pp. 965968.
differential amplier with and without slew rate improvement [9] S.-K. Shin, et al., A slew rate controlled output driver using PLL as compensa-
architecture is shown in Fig. 7 for an input signal with a rise and tion circuit, in: Proceedings of IEEE ESSCIRC, 2002, pp. 207210.
[10] A.P. Perez, Y.B. Nithin Kumar, E. Bonizzoni, F. Maloberti, Slew-rate and gain
fall time of 50 ps. enhancement in two stage operational ampliers, in: IEEE International
The above results clearly show that the stability of the Symposium on Circuits and Systems (ISCAS), 2009, pp. 24852488.
differential amplier with CMFB circuit does not get affected [11] S. Yin, Y. Sun, X. Li, Digital 1 V 82 mW pseudo-two-stage class-AB OTA,
Tsinghua Science and Technology 14 (5) (2009) 601605.
due to the presence of the slew rate improvement architecture. [12] S.C. dela Cruz, M.G.T. delos Reyes, T.C. Gaffud, T.V.F. Abaya, M.T.A. Gusad,
It may be noted that the increase in power dissipation in a M.D. Rosales, Design and implementation of operational ampliers with
circuit with slew rate enhancement architecture come from the programmable characteristics in a 90 nm CMOS process, in: European
Conference on Circuit Theory and Design (ECCTD), 2009, pp. 209212.
power dissipated in SR enhancement circuitry. Detect and control
[13] J. Ramrez-Angulo, R.G. Carvajal, A. Torralba, C. Nieva, A new Class-AB
circuit (SR enhancement circuitry) can be shared among several differential input stage for implementation of low-voltage highslew rate
op-amps in applications, where input of all op-amps is connected op-amps and linear transconductors, in: Proceedings of IEEE International
Symposium Circuits and Systems, ISCAS, vol. 1, May 2001, pp. 671674,
to similar signals. Therefore, the power dissipation in slew rate
doi:10.1109/ISCAS.2001.921945.
enhancement circuitry will be shared among multiple op-amps [14] F. Moraveji, A wide-band, low-power, high slew rate voltage feedback
and the percentage increase in power dissipation will be further operational amplier, IEEE Journal of Solid-State Circuits 31 (1) (1996) 1016.
reduced. The sharing of detect and control parts of the slew rate [15] N. Charalampidis, K. Hayatleh, B.L. Hart, F.J. Lidgey, A wide bandwidth
voltage-follower with low distortion and high slew rate, in: Proceedings of
enhancement circuitry among multiple op-amps will be covered the 13th IEEE International Conference on Electronics, Circuits and Systems,
in a separate work. (ICECS), December 2006, pp. 256259. doi:10.1109/ICECS.2006.379774.
R.A. Thakker et al. / Microelectronics Journal 42 (2011) 758765 765

[16] J.T. Sundby, CMOS opamp with large sinking and sourcing currents and high circuits and SRAM, in: IEEE Symposium on VLSI Technology, 1214 June
slew rate, U.S. Patent No. 5325069, June 1994. 2007, pp. 106107.
[17] M. Miura, Operational amplier with self control circuit for realizing high slew [23] R.A. Thakker, C. Sathe, A.B. Sachid, M.S. Baghini, V. Ramgopal Rao, M.B. Patil,
rate throughout full operating range, U.S. Patent No. 6987420 B2, January 2006. A novel table-based approach for design of FinFET circuits, IEEE Transactions
[18] K. Feng-Sung, Slew rate enhancement circuit via dynamic output stage, U.S. on Computer-Aided Design of Integrated Circuits and Systems (July) (2009)
Patent No. 7164298 B2, January 2007. 10611070.
[19] C.-H. Chuang, T.-K. Tseng, H.-C. Jiang, Slew-rate control circuitry with output [24] M.B. Patil, A new public-domain program for mixed-signal simulation, IEEE
buffer and feedback, U.S. Patent No. 7652511 B2, January 2010. Transactions on Education 45 (2) (2002) 187193.
[20] Synopsys Sentaurus Design Suite [Online]. /http://www.synopsys.comS. [25] R.A. Thakker, C. Sathe, A.B. Sachid, M.S. Baghini, V. Ramgopal Rao, M.B. Patil,
[21] M. Srivastava, B. Verma, M.S. Baghini, C. Russ, D.K. Sharma, H. Gossner, Automated design and optimization of circuits in emerging technologies, in:
V. Ramgopal Rao, Benchmarking the device performance at sub 22 nm node Proceedings of the 14th Asia and South Pacic Design Automation Confer-
technologies using an SoC framework, in: IEEE Electron Devices Meeting, 79 ence, January 2009, pp. 504509.
December 2009, pp. 505508. [26] M. Srivastava, M.S. Baghini, D.K. Sharma, V. Ramgopal Rao, A novel and
[22] K. von Arnim, et al., A low-power multi-gate FET CMOS technology with robust approach for common mode feedback using IDDG FinFET, IEEE
13.9 ps inverter delay, large-scale integrated high performance digital Transactions on Electron Devices 55 (11) (2008) 32743282.