Академический Документы
Профессиональный Документы
Культура Документы
I. INTRODUCTION
Authorized licensed use limited to: NATIONAL INSTITUTE OF TECHNOLOGY CALICUT. Downloaded on March 08,2020 at 08:31:40 UTC from IEEE Xplore. Restrictions apply.
132 IEEE TRANSACTIONS ON CIRCUITS AND SYSTEMS—II: EXPRESS BRIEFS, VOL. 54, NO. 2, FEBRUARY 2007
B. CMOTA
Another large-swing amplifier topology that avoids the
compensation capacitors is the current-mirror OTA (CMOTA),
which is depicted in Fig. 1(b). Because of the low impedance
seen at the drain of the diode-connected transistor, , the
associated pole frequency of the first stage is fairly far and
has little effect on the amplifier phase margin and bandwidth.
Therefore, frequency compensation is not required. For the
same design conditions as the previous two-stage OTA, the
current consumed by a CMOTA can be expressed as [2]
(2) Fig. 4. Current-shunt CMOTA with current reuse and body inputs.
where is the size ratio between mirroring devices, and grades [4]. In addition, the nondominant pole frequency is low-
. Comparing with (1), eliminating the compensation ca- ered. This effect will be discussed in Section IV.
pacitors results in considerable power reduction. However, the
III. PROPOSED OTA ARCHITECTURE
major drawback of a CMOTA is its low voltage gain.
A. CMOTA With Current Reuse
C. Current-Shunt CMOTA
In Fig. 2, the currents flowing into ground can serve mul-
The voltage gain of a CMOTA can be improved by raising
tiple functions. In fact, it can be utilized as the second input stage
the circuit output impedance. In addition to cascoding transis-
of the OTA. As shown in Fig. 3, the two shunting current sources
tors, using longer channel devices or lowering the transistor bias
are replaced with the circuit encircled by the dotted line. Tran-
current also boosts the impedance. In [3], a circuit technique is
sistors form the second input pair, where provide
employed to lower the bias current of the output branch. As de-
necessary level shifting for low-voltage operation. The tail cur-
picted in Fig. 2, by shunting partial bias current ( , )
rent of the second input pair is to maintain the same level of
of the first stage to ground, less dc current is mirrored to the
current shunting as in Fig. 2. If the transconductance is
output branches, while the ac signal current is unaffected. The
designed to be roughly equal to , the current-reuse tech-
output bias current is now . This method enhances
nique provides an additional 6-dB gain with small current in-
the output impedance, and hence the voltage gain, by
crease (two bias currents for level shifters).
times. The voltage gain of a current-shunt CMOTA is
formulated as B. Body Inputs
In Fig. 3, the body terminals of are connected to their
source nodes to avoid the body effect. Since a body terminal
can also behave as a second gate [5]–[7], we can apply the
level-shifted inputs (at nodes and ) to the body terminals
(3) of , as shown in the dashed lines of Fig. 4. Since the dc
voltages of nodes and are comparable to
and are the transistor output resistances, and the those of the node , the body-source junctions of will
and are the channel-length modulation coefficients not enter the forward-biased region. The body-drain transcon-
for transistors and . For a fixed current-mirror ductance of the pMOS transistors contributes to the
ratio , as increases, the gain improves but the slew rate de- overall and thereby increases the voltage gain.
Authorized licensed use limited to: NATIONAL INSTITUTE OF TECHNOLOGY CALICUT. Downloaded on March 08,2020 at 08:31:40 UTC from IEEE Xplore. Restrictions apply.
LIN et al.: A 0.8-V 0.25-mW CMOTA 133
The body terminals of can also serve as inputs, if a major parasitic poles are located at nodes and in Fig. 5. The
deep-Nwell process option is available. These body nodes are pole frequencies associated with these two nodes are
connected to and , as shown in the dotted lines of Fig. 4.
The dc levels of and are below , (6)
which are comparable to those of the source voltages of tran-
sistors . Hence, the body-source junctions of will (7)
not conduct.
and represent the total capacitance at nodes and ,
C. Overall Proposed OTA
respectively. These frequencies are typically designed to be at
The complete proposed current-mirror-based OTA is shown least three times the GBW frequency [10]. As approaches
in Fig. 5. With a push-pull output stage [8], [9], the voltage 1, the OTA achieves a higher gain. However, the transconduc-
gain is further increased (by roughly a factor of 2 in this work). tances of and ( and ) decrease, so are
This topology also makes the biasing of the output stage some- and , which degrades the frequency response. There-
what simpler. The common-mode feedback (CMFB) network fore, these design parameters must be selected with care.
consists of resistors , error amplifier cm , and transistors Compared with the current-shunt approach, the current-reuse
. Resistors are chosen to be large k to technique not only increases the effective input trans-conduc-
avoid degrading the OTA gain. The error amplifier also employs tance, but it may also be advantageous in lowering the parasitic
the current-shunt technique to enhance its voltage gain. capacitance at node . The reason is that the transistors
are usually designed with minimum channel length to maxi-
IV. CIRCUIT ANALYSIS mize the transconductance; while in the case of current sources
, longer channel length is typically required to minimize the
A. Voltage Gain and GBW channel-length modulation effect.
The total voltage gain of the proposed OTA of Fig. 5
can be expressed as
B. Noise Analysis
Authorized licensed use limited to: NATIONAL INSTITUTE OF TECHNOLOGY CALICUT. Downloaded on March 08,2020 at 08:31:40 UTC from IEEE Xplore. Restrictions apply.
134 IEEE TRANSACTIONS ON CIRCUITS AND SYSTEMS—II: EXPRESS BRIEFS, VOL. 54, NO. 2, FEBRUARY 2007
(9)
Fig. 6. Simplified transfer characteristic model of the proposed OTA.
Authorized licensed use limited to: NATIONAL INSTITUTE OF TECHNOLOGY CALICUT. Downloaded on March 08,2020 at 08:31:40 UTC from IEEE Xplore. Restrictions apply.
LIN et al.: A 0.8-V 0.25-mW CMOTA 135
VI. CONCLUSION
A low-voltage low-power CMOS OTA is introduced in
this brief. The proposed circuit is based on the current-mirror
topology. To circumvent the low-gain problem of a CMOTA,
several design techniques are employed. First, the output
impedance is increased by reducing the bias currents of the
output branches, which is realized by shunting partial mirror
Fig. 8. Step response settling behavior waveforms.
currents away. These shunt currents are then reused to realize
the second input stage. In addition, body terminals are uti-
TABLE I lized as inputs to augment the transconductance. Finally, the
PERFORMANCE SUMMARIES OF THE PROPOSED OTA AND COMPARISON
WITH OTHER LOW-VOLTAGE OTA DESIGNS push-pull output stage further enhances the gain. The criteria
for determining the circuit bandwidth and gain parameters are
discussed. The noise and settling behavior of the proposed
circuit are also analyzed in this brief. It is identified that the
level shifter contributes considerable noise and adds a PZP to
the OTA transfer function, which prolongs the circuit settling
time. These level shifters are required for a low-voltage (0.8 V)
operation. In the case where supply voltage requirement is
relaxed, the level shifters can be eliminated. Thus, the noise and
settling behavior of the proposed OTA can be further improved.
REFERENCES
[1] D. A. Jones and K. Martin, Analog Integrated Circuit Design. New
York: Wiley, 1997, ch. 5.
[2] L. Yao, M. Steyaert, and W. Sansen, “A 1-V 140-W 88-dB audio
sigma-delta modulator in 90-nm CMOS,” IEEE J. Solid-State Circuits,
vol. 39, no. 11, pp. 1809–1818, Nov. 2004.
[3] ——, “A 0.8-V, 8-W, CMOS OTA with 50-dB gain and 1.2-MHz
GBW in 18-pF load,” in Proc. Eur. Solid-State Circuits Conf., Sep.
2003, pp. 297–300.
The simulated open-loop gain and phase response is shown in [4] A. J. Lopez-Martin, S. Baswa, J. Ramirez-Angulo, and R. G. Carvajal,
Fig. 7. The additional PZP contributed by the level shifters can “Low-voltage super class-AB CMOS OTA cells with very high slew
be observed in Fig. 7. To examine the effect of this doublet on rate and power efficiency,” IEEE J. Solid-State Circuits, vol. 40, no. 5,
pp. 1068–1077, May 2005.
the circuit settling behavior, the OTA is configured as a closed- [5] A. Guzinski, M. Bialko, and J. C. Matheau, “Body driven differential
loop unity-gain amplifier, and a 10-mV step is applied to the amplifier for application in continuous-time active-C filter,” in Proc.
amplifier input. Fig. 8 shows the simulated output waveforms. ECCD, 1987, pp. 315–319.
[6] B. J. Blalock, P. E. Allen, and G. A. Rincon-Mora, “Designing 1-V op
Even though the unity-gain amplifier has a 3-dB frequency amps using standard digital CMOS technology,” IEEE Trans. Circuits
over 160 MHz, the circuit still takes about 10.9 ns Syst. II, Anal. Digit. Signal Process., vol. 45, pp. 769–780, Jul. 1998.
to settle to within 1% accuracy. The settling time exceeds the [7] S. Chatterjee, Y. Tsividis, and P. Kinget, “0.5-V analog circuit tech-
niques and their application in OTA and filter design,” IEEE J. Solid-
value of predicted by the simple linear State Circuits, vol. 40, no. 12, pp. 2373–2387, Dec. 2005.
model [1] and shows the slow-settling behavior. To observe the [8] S. Rabii and B. A. Wooley, “A 1.8-V digital-audio sigma-delta mod-
effect of the level shifters on the settling time, a comparison ulator in 0.8-m CMOS,” IEEE J. Solid-State Circuits, vol. 32, no. 6,
pp. 783–796, Jun. 1997.
is performed where the level shifters in the OTA are replaced [9] J. Ramírez-Angelo, M.-S. Sawant, S. Thoutam, A. J. López-Martin,
with ideal behavioral models (hence eliminating the PZP). The and R. G. Carvajal, “New low-voltage class AB/AB CMOS opamp
simulated waveform is also plotted in Fig. 8. The 1% settling with rail-to-rail input/output swing,” IEEE Trans. Circuits Syst. II, Exp.
Briefs, vol. 53, no. 4, pp. 289–293, Apr. 2006.
time is reduced to 7.3 ns. These results indicate the PZP plays [10] K. R. Larker and W. M. Sansen, Design of Analog Integrated Circuits
an important role in the settling behavior [12]. and Systems. New York: McGraw-Hill, 1994, ch. 6.
The proposed OTA employs body-input topology which gives [11] B. Razavi, Design of Analog CMOS Integrated Circuits. New York:
McGrill-Hill, 2000, ch. 6.
rise to small input current through the reversed-biased junctions. [12] B. Kamath, R. Meyer, and P. Gray, “Relationship between frequency
The leakage current in general rises with temperature. The sim- response and settling time of operation amplifiers,” IEEE J. Solid-State
ulated value is less than 10 pA under worst corner. The nonzero Circuits, vol. SC-9, no. 12, pp. 347–352, Dec. 1974.
[13] J. Rosenfeld, M. Kozak, and E. G. Friedman, “A bulk-driven CMOS
input current results in an input resistance on the order of 10 G . OTA with 68-dB dc gain,” in Proc. IEEE Int. Electron., Circuits Syst.
The effective input capacitance is estimated at about 0.4 pF. The (ICECS), Dec. 2004, pp. 5–8.
Authorized licensed use limited to: NATIONAL INSTITUTE OF TECHNOLOGY CALICUT. Downloaded on March 08,2020 at 08:31:40 UTC from IEEE Xplore. Restrictions apply.