XU AND ABIDI: ANALYSIS AND DESIGN OF REGENERATIVE COMPARATORS FOR LOW OFFSET AND NOISE 3
originally on CC . When accounted for algebraically, charge 1) C L 2CC . For a dynamic amplifier with large gain,
is still conserved since the two capacitors are connected by a s p T p is large and negative, so the exponential in (17) is
two-terminal device; and also energy, after taking into account negligible.
that both capacitors were pre-charged to the supply voltage. 2) C L ∼ CC . This is rare. Here s p T p is very small,
To determine whether the charge transfer is over a stable so e+s p T p 1 + s p T p in (17), and (C L − CC ) cancels a
or a regenerative transient, we examine the sign of the real term in the numerator.
pole that characterizes the loop formed by CC , C L and the 3) C L 2CC . This might be true if a large number of cal-
controlled source. It is sufficient for a check on stability to ibration capacitors are attached at CC . This will reverse
assume that the controlled source is linear. Then this pole is the sign of s p T p in (18), so a growing exponential
C L − CC term will dominate the parentheses in (17). The sign
s p = −gm3,4 (12) of VO D (Ts + T p ) will not change.
C L CC The final voltage then simplifies to:
When C L > CC , the pole is a negative real quantity and ⎧
the charge transfer will decay to a well-defined final value. ⎪ Ts + T p
⎪
⎪ if CC C L /2
But when the pole is positive, the charge transfer will grow ⎪
⎪ C −C
⎪
⎨ T L+ 1 TC
exponentially for the duration that the FET behaves as a linear
v O D (Ts + T p ) = i I D ×
s 2 p Tp
element: when it enters into a different regime such as triode · if CC ≈ C L
⎪
⎪ C L − CC τ p
region, regeneration will be clamped. We are interested in ⎪
⎪
⎪
⎪ τ p − Ts
the voltage on C L at the end of the time window T p , when ⎩ · e+s p T p if CC 2C L
propagation concludes and regeneration starts. Then, even a C L − CC
growing exponential waveform will deposit a well-defined (19)
voltage on C L . Voltage gain ( A V ) is found by using i I D = gm1,2 v I D to obtain
Thus there are two contributions to VO D . First there is v O D /v I D
the charge integrated on CC that redistributes to C L through ⎧
charge transfer that we have just described. As well, there is ⎪ C L + 2CC 2Vt
⎪
⎪ · if CC C L /2
the fraction of i I D through M3, M4 that integrates on C L . ⎪
⎪ C − C V
⎪ L
⎨ C + 3C C D S AT 1,2
We will calculate these two components below using the L C Vt2
transfer function, AV = · if CC ≈ C L
⎪
⎪
⎪ 2V C+CV
⎪
V D S AT 1,2 V D S AT 3,4
2V
1 ⎪
⎪ t D S AT 3,4 t
⎩ · exp if CC 2C L
VO D (s) = I I D (s) s (13) V D S AT 1,2 V D S AT 3,4
s(C L − CC ) 1 −
sp (20)
From (2), the charge on CC at the end of the sampling With typical values, we may conclude that in all reasonable
phase is cases the voltage gain VO (Ts + T p )/v id ≈ 5 ∼ 10. It covers
CC Vt N all cases of C L /CC ; and gives a meaningful expression when
Q C (Ts ) = CC VC D (Ts ) = 2 vI D (14) CC C L , in contrast to [13, Eq. (5)] which implies that
V D S AT 1,2 preamplification gain would then be infinite. A useful way
In propagation phase this charge defines an impulse current of understanding it is that with this capacitive inequality,
i 1 (t) = Q C (Ts )δ(t − Ts ) that is applied to the zero-state the source degeneration on the cross-coupled pair M3,4 is
network whose transfer function is (13). After the inverse lowered enough that the pair develops sufficient loop gain to
transform, regenerate, but that the regenerating waveform is sampled at
Q C (Ts ) the end of a prescribed time window.
v O D1 (Ts + T p ) = 1 − e+s p T p (15) 4) Regeneration Phase: At the end of the propagation
C L − CC phase, the cross-coupled PMOS pair, M5, M6, turns ON and,
Meanwhile, a fraction of i id keeps integrates over the period since its sources are connected together, it always regenerates
T p and adds to VO the voltage the voltage (19) at the output nodes (Fig.6(d)). At start of
regeneration the PMOS pair is governed by a pole that lies in
i id · T p i id · τ p
the right-half s-plane sreg = +gm5 /C L .
v O D2 (Ts + T p ) = − 1 − e+s p T p (16)
C L − CC C L − CC
C. FET Mismatch
The net output is the algebraic sum of the two Estimation of the net offset arising from FET mismatches
T + T − τ τ p − Ts requires knowledge of the transfer functions from each mis-
· e+s p T p
s p p
v O D (Ts + T p ) = i id + match source to the comparator output. Shown in Fig.7 are the
C L − CC C L − CC
equivalent differential mode circuits during sampling and prop-
(17)
agation phases, including independent current sources mod-
To calculate the term in parentheses we must know the elling mismatch. These circuits apply while M1∼M4 operate
magnitude and sign of in saturation.
These circuits show that Vt and β mismatch in the input
2|Vt P | C L − CC pair M1, M2 appear at the comparator input as an effective
s p Tp = · (18)
V D S AT 3,4 CC differential offset voltage.
Consider three cases, where the factor of two in the inequalities I M1,2 β1,2 V D S AT 1,2
is to indicate an approximate boundary between the three v os | M1,2 = = Vt 1,2 + · (21)
gm1,2 β1,2 2
cases.
This article has been accepted for inclusion in a future issue of this journal. Content is final as presented, with the exception of pagination.
XU AND ABIDI: ANALYSIS AND DESIGN OF REGENERATIVE COMPARATORS FOR LOW OFFSET AND NOISE 5
D. Capacitor Mismatch
CC and C L , both grounded capacitors, are also subject
to mismatch. Indeed following (8) we advocate their delib-
erate unbalance to compensate FET mismatches. Capacitor Fig. 9. Comparison between theoretical calculation and simulation of
mismatch is modelled by a differential current (Fig.7(b)). dynamic offset introduced by CC : (a) Vary CC ; (b) Vary CC ; (c) Vary
V DS AT 1,2 ; (d) Vary V D D .
Through the sampling and propagation phases, the common-
mode to differential-mode coupling current due to CC mis-
match integrates charge onto CC and C L . The total charge common-mode voltage excursion of Vt N + |Vt P |, which is
induced by CC mismatch is CC (Vt N + |Vt P |). The input- almost twice the excursion of the voltage on C L . Eqs. (22)
referred offset that results is the differential input voltage and (23) show that the offsets introduced by CC , C L are,
(v os = Vin1 − Vin2 ) that will counter this mismatch by driving contrary to the expressions in [14], both independent of V D D .
the differential voltage across C L to zero. From (10) and (11) This is important for design.
Fig. 8 and Fig. 9 compare the simulated dynamic offsets
gm1,2 v os × (Ts + T p ) = CC (Vt N + |Vt P |) against predictions from (22) and (23) in a StrongArm latch
2CC V D S AT 1,2 designed in 28 nm FD-SOI CMOS (FET sizes in Fig. 20).
⇒ v os · (22) The predictions match simulated results well across different
C L + 2CC 2 conditions. The input-referred offset remains independent of
Mismatch in CC appears through both sampling and prop- V D D until it is as low as 850 mV, below which there is no
agation phase, whereas mismatch in C L matters only during longer the voltage headroom required to maintain M1, M2 in
propagation phase. The input-referred offset voltage due to C L saturation.
mismatch must counter the charge injected by C L into CC
and C L (Fig. 7(b)). Thus E. Offset Compensation
With this background, we can explain satisfactorily how
gm1,2 v os × (Ts + T p ) = C L |Vt P | the offset calibration strategy first proposed in [15] and now
C L V D S AT 1,2 widely used works in latched comparators. We have estab-
⇒ v os · (23) lished that mismatch in the threshold voltage and β of the pair
C L + 2CC 2
M1, M2 will dominate. Now if the capacitances CC and C L
Comparing (22) with (23), we see that CC unbalance induces are fine tuned in closed-loop under digital control to create an
twice the offset of C L imbalance because CC undergoes a almost equal offset as given by (22) and (23) but opposite
This article has been accepted for inclusion in a future issue of this journal. Content is final as presented, with the exception of pagination.
XU AND ABIDI: ANALYSIS AND DESIGN OF REGENERATIVE COMPARATORS FOR LOW OFFSET AND NOISE 7
Fig. 12. (a) The original double-tail latch from [20]; (b) The improved double-tail latch from [21].
the two comparator inputs are themselves unequal, the pulse of Initially these FETs are biased into deep triode by the large
common-mode input current converts into a larger differential voltage V P1 , V P2 applied to their gates. Their β is much
voltage. For example, in flash ADCs one input may be larger than of M3, M4 to suppress any inclination of the
connected to a resistor reference ladder, which is modelled by cross-connected pair M5, M6 to regenerate. As V P1 , V P2 ramp
a distributed RC circuit. The input current impulse, referred down, MP1, MP2 amplify the differential input; at some point
to as ‘kickback’ [24], can induce transients on all tap voltages on this ramp M1a, M2a release their hold on M3, M4, enabling
[22, Ch. 8]. This produces a signal-dependent reference error. the latch M3∼M6 to regenerate. It is intended that the sign of
When all comparators share the same reference ladder and the amplified differential voltage coupled into the latch through
its RC settling is slow, kickback from one decision interferes M1a, M2a, will determine the regenerated binary output. Only
with the next decision. the preamplifier is clocked in this circuit; everything else
Another kickback appears during regeneration, when the follows by self-timing, concluding in regeneration of the latch
latching of the cross-coupled inverters M3∼M6 perturbs the circuit.
sources driving the comparator input through capacitive cou- Reference [26] points out with simulations that this arrange-
pling via M1, M2. This effect is different in its origin from ment performs poorly. We will explain why. Consider the
the common-mode kickback and its effect can sometimes be dynamics of the coupling FETs M1a, M2a. They turn on in
quite benign. In the StrongArm latch this kickback appears deep triode with a large VG = V D D and V D ≈ 0. As VG
when regeneration is almost completed and therefore cannot ramps down, their V D rises, until VG crosses below Vt 0 to
interfere with that decision. Charge-redistribution SAR ADCs, turn them off. This describes a time trajectory on the FET
where one comparator makes every decision, are a special I D -V D S plane which starts in deep triode when M1a, M2a’s gm
case. [25, Appendix D-D] shows that any differential-mode is low; then they enter saturation, when V D rises to 1/2V D D
kickback causes no harm as long as the comparator is reset but at that point VG → Vt 0 , so again their gm is low. Over
before every decision. this trajectory the transconductance is weak for most of the
time, so M1a, M2a are ineffective in coupling the preamplified
IV. D OUBLE TAIL C OMPARATOR voltage at their gates into the latch. Offset and noise in the
The “double tail” comparator preserves the benefits of latch M3∼M6 may now have a pronounced effect.
internal dynamic amplification in the StrongArm circuit at
low supply voltages, or when the input common-mode bias is A. Details of Operation
unsuitable. Originally developed to buffer the driving circuit
from charge kickback during regeneration [20], its circuit Reference [26] goes on to offer a better circuit (Fig. 12(b)).
structure (Fig. 12(a)) also enforces zero static power dissi- We recognize this as a dynamic amplifier MP1, MP2 driving
pation. A dynamic preamplifier precedes a latch in voltage a StrongArm latch in voltage cascade. Why does this not
cascade, that is, in shunt.2 When CLK goes high, the differ- suffer the same fate as the circuit of Fig. 12(a)? It has to
ential input is amplified and appears superposed on the output do with the direction of the ramping bias that the comple-
common-mode (bias) voltage that, in this circuit, ramps down mentary (folded) amplifier applies to the StrongArm latch.
from an initial voltage of V D D to 0. This changing common- The preamplifier drives the input pair M1a, M2a of the
mode voltage will eventually drive the preamplifier differential StrongArm latch with a positive-going ramp, on which is
pair MP1, MP2 into triode, which will then discharge the superposed the continuously growing differential signal. This
amplified differential voltage stored on capacitors C pre . Then ramp turns on the input pair of the StrongArm latch and softly
the preamplifier will shut off its own bias current. The pro- releases the reset switches M1b, M2b, M1c, M2c. M1a, M2a’s
gressively amplified input is available over the time window transconductance rises with the ramp because the pair is in
between when the clock turns on, and before the amplifier saturation, while its differential input grows continuously. The
shuts itself off and erases the amplified voltage. StrongArm goes through its self-timed phases of sampling,
The FETs M1a, M2a couple the dynamic preamplifier’s propagation, and regeneration. Since everything operates as
output to a static latch. They are key to correct operation. expected, the dynamic amplification is strong and the com-
parator’s input-referred offset and noise will now be deter-
2 As opposed to the StrongArm latch, where the dynamic amplifier is in mined almost entirely by the preamplifier’s input pair MP1,
series cascade. MP2.
This article has been accepted for inclusion in a future issue of this journal. Content is final as presented, with the exception of pagination.
XU AND ABIDI: ANALYSIS AND DESIGN OF REGENERATIVE COMPARATORS FOR LOW OFFSET AND NOISE 9
NBW of 1/ T pre + 23 (Ts + T p ) [Hz] . Thus the input-referred
mean-square noise voltage is
4kTγ 1
v id
2
= · (32)
gmp1,2 T pre + 23 (Ts + T p )
XU AND ABIDI: ANALYSIS AND DESIGN OF REGENERATIVE COMPARATORS FOR LOW OFFSET AND NOISE 11
TABLE I
M EASURED RMS O FFSET VS . C ALCULATED . FET M ISMATCH
PARAMETERS : A V t = 3.5mV·μm, Aβ = 2.5%·μm
TABLE II
M EASURED RMS O FFSET VS . C ALCULATED . FET M ISMATCH
PARAMETERS : A V t = 2.3mV·μm, Aβ = 1.5%·μm
Fig. 20. (a) StrongARM comparator with calibration capacitors. All dimen-
sions in μm, V D D =1 V; (b) Offsets, noise, latency and regeneration constant
before/after calibration.
Fig. 19. (a) Test circuit in [17]. All dimensions in μm, V D D =1 V;
(b) Calculated input-referred noise versus simulation results in [17]. The 1 V supply is adequate for proper operation of a Stron-
gArm latch, which we chose (Fig. 20(a)) for its higher noise
power consumption. First, increasing W/L of MP1, MP2√by filtering efficiency as discussed in Sec. IV-B. Its input common
2× lowers the dominant source of mismatch right away by 2. mode voltage is set as 550 mV for the optimum trade-off
Second, raising R P in the preamplifier from 3 k to 8 k between noise and speed, as described in Sec. V. Sizing of
boosts the preamplifier gain, but also lowers the preamplifier’s the FETs in the StrongArm latch is straightforward.
output bias from 900mV to 650mV, restoring the StrongArm’s The loading on the comparator output is determined by the
dynamic voltage gain to about 10× and suppressing the offset logic gates realizing the successive approximation algorithm.
contributed by M3, M4. The higher voltage gain also shields The size of the input differential pair is limited by the ability
the StrongArm latch from the hysteresis effect. Table II shows of the previous stage to drive it. The remaining FETs in
the predicted improvement with this remedy. Simulations the StrongArm latch are sized for fast regeneration with no
verify quantitatively that the offsets are lowered to an extent regard to offsets (Fig.20(a)), which will be suppressed with
close to what analysis predicts. calibration capacitors. The next step in design is to decide
how to allocate calibration capacitors between CC and C L to
C. Input-Referred Noise of StrongArm Latch cover process spreads.
Reference [17] reports on simulated input-referred RMS Without calibration, σ of the comparator offset is 8 mV,
noise of a StrongArm latch versus the input common mode dominated by the input differential pair. Using (22) and (23)
voltage as gm /I of the input differential pair changes for calculation, we decided to place most of the calibration
Fig. 19(a). Using our (29), we can predict this noise accurately capacitance at CC for coarse offset calibration that covers
as shown in Fig. 19(b). However, the predicted noise deviates the expected spreads in FET mismatch Fig.20(b). In total,
from the simulated value at an input common mode voltage 6 bits of binary-weighted MOS capacitors are used for coarse
of more than 600 mV. This is because of the increasing noise calibration, with an MSB capacitor of 16 fF. This leaves 5 bits
contribution from M3, M4 as M1, M2 are pushed deeper for fine calibration, entailing a total capacitance of 2.4 fF. This
into triode; the calculations are left out for brevity. This array is placed at C L , adding to the fixed load of 10 fF. The
underscores once again the importance in the StrongArm regeneration time constant τ and therefore metastable error
circuit of choosing the correct common mode voltage in order rate are not much affected, but the RMS offset after calibration
to avail internal dynamic amplification. is tightened to ±0.4 mV (Fig. 20(b)).
The total calibration capacitance at CC and C L is deter-
VII. D ESIGN G UIDE mined by specifications on thermal noise from (28) and (29).
Since offset can be suppressed with calibration capacitors These expressions offer straightforward estimation of noise
at relatively low overhead except for a small penalty in speed, without resort to time-consuming transient noise simulations
the focus of comparator design shifts to lowering noise. Noise of regenerative comparators and iterative optimization. Includ-
reduction is expensive as it usually requires greater dissipation ing the calibration capacitors, the input-referred noise is
of power, yet it is critical in applications such as ADCs without lowered to 0.35 mV RMS.
residue amplification or in decision feedback equalizers. The process of designing the optimum StrongArm com-
We have designed the comparators in a 2.5 GS/s 10 bit parator is now clear. Transistors are first sized to maximize
time-interleaved charge redistribution SAR ADC, fabricated speed given the load constraints imposed by the preceding and
in 28 nm FD-SOI process. With 1 LSB=1.5 mV, considerations subsequent circuits. Calibration capacitors are then applied for
of comparator noise drove the trade-off in power and speed. both offset calibration and noise filtering. The key idea in noise
This article has been accepted for inclusion in a future issue of this journal. Content is final as presented, with the exception of pagination.
XU AND ABIDI: ANALYSIS AND DESIGN OF REGENERATIVE COMPARATORS FOR LOW OFFSET AND NOISE 13
The output noise variance v 2O of the system sampled at time [20] D. Schinkel, E. Mensink, E. Klumperink, E. van Tuijl, and B. Nauta,
t0 is “A double-tail latch-type voltage sense amplifier with 18ps setup+hold
time,” in IEEE Int. Solid-State Circuits Conf., Dig. Tech. Papers,
Sin +∞ 2 San Francisco, CA, USA, Feb. 2007, pp. 314–605.
v O =
2
h (t0 , τ )dτ (42) [21] M. Miyahara and A. Matsuzawa, “A low-offset latched comparator
2 −∞ using zero-static power dynamic offset cancellation technique,” in Proc.
IEEE Asian Solid-State Circuits Conf., Taipei, Taiwan, Nov. 2009,
The time varying impulse response of the circuit in Fig.15(c) is pp. 233–236.
t [22] M. J. M. Pelgrom, Analog-to-Digital Conversion, 3rd ed. Dordrecht,
1 β1 I 0 1
h(t, τ ) = VO (t) = ·t
− C
dt
The Netherlands: Springer, 2016.
τ C pre C pre C [23] H. Huang, H. Xu, B. Elies, and Y. Chiu, “A non-interleaved 12-b
L C
330-MS/s pipelined-SAR ADC with PVT-stabilized dynamic amplifier
gm1 (t ) achieving sub-1-dB SNDR variation,” IEEE J. Solid-State Circuits,
vol. 52, no. 12, pp. 3235–3247, Dec. 2017.
β1 I 0 1
= · (t 2 − τ 2 ) (43) [24] P. M. Figueiredo and J. C. Vital, “Kickback noise reduction techniques
C pre (C L
− CC
) 2 for CMOS latched comparators,” IEEE Trans. Circuits Syst., II, Exp.
Briefs, vol. 53, no. 7, pp. 541–545, Jul. 2006.
The voltage gain and input referred noise are then obtained [25] T. Iizuka, T. Ito, and A. A. Abidi, “Comprehensive analysis of distortion
using (43). in the passive FET sample-and-hold circuit,” IEEE Trans. Circuits Syst.
I, Reg. Papers, vol. 65, no. 4, pp. 1157–1173, Apr. 2018.
[26] M. Miyahara, Y. Asada, D. Paik, and A. Matsuzawa, “A low-noise
R EFERENCES self-calibrating dynamic comparator for high-speed ADCs,” in Proc.
[1] P. E. Allen and D. R. Holberg, CMOS Analog Circuit Design, 3rd ed. IEEE Asian Solid-State Circuits Conf., Fukuoka, Japan, Nov. 2008,
New York, NY, USA: Univ. Oxford, 2012. pp. 269–272.
[27] R. J. Schwarz and B. Friedland, Linear Systems. New York, NY, USA:
[2] A. Abidi and H. Xu, “Understanding the regenerative comparator cir-
McGraw-Hill, 1965.
cuit,” in Proc. IEEE Custom Integr. Circuits Conf., San Jose, CA, USA, [28] C.-C. Liu, S.-J. Chang, G.-Y. Huang, and Y.-Z. Lin, “A 10-bit 50-MS/s
Sep. 2014, pp. 1–8. SAR ADC with a monotonic capacitor switching procedure,” IEEE J.
[3] E. A. Vittoz, “Dynamic analog techniques,” in Design of Analog- Solid-State Circuits, vol. 45, no. 4, pp. 731–740, Apr. 2010.
Digital VLSI Circuits for Telecommunications and Signal Processing, [29] B. Verbruggen, M. Iriguchi, and J. Craninckx, “A 1.7 mW 11b 250 MS/s
2nd ed. J. E. Franca and Y. Tsividis, Eds. Englewood Cliffs, NJ, USA: 2-times interleaved fully dynamic pipelined SAR ADC in 40 nm digital
Prentice-Hall, 1994, ch. 6. CMOS,” IEEE J Solid-State Circuits, vol. 47, no. 12, pp. 2880–2887,
[4] R. D. Middlebrook, Differential Models. New York, NY, USA: Wiley, Dec. 2012.
1963. [30] H. S. Bindra, C. E. Lokin, D. Schinkel, A.-J. Annema, and B. Nauta,
[5] T. H. O’Dell, Circuits for Electronic Instrmntatn. Cambridge, U.K.: “A 1.2-V dynamic bias latch-type comparator in 65-nm CMOS with
Cambridge Univ. Press, 1991. 0.4-mV input noise,” IEEE J. Solid-State Circuits, vol. 53, no. 7,
[6] J. He, S. Zhan, D. Chen, and R. L. Geiger, “Analyses of static and pp. 1902–1912, Jul. 2018.
dynamic random offset voltages in dynamic comparators,” IEEE Trans. [31] P. R. Kinget, “Device mismatch and tradeoffs in the design of analog
Circuits Syst. I, Reg. Papers, vol. 56, no. 5, pp. 911–919, May 2009. circuits,” IEEE J. Solid-State Circuits, vol. 40, no. 6, pp. 1212–1224,
[7] H. Taub, D. L. Schilling, and G. Saha, Principles of Communication Jun. 2005.
Systems, 2nd ed. New York, NY, USA: McGraw-Hill, 1986. [32] L. L. Lewyn, T. Ytterdal, C. Wulff, and K. Martin, “Analog circuit
[8] T. Kobayashi, K. Nogami, T. Shirotori, and Y. Fujimoto, “A current- design in nanoscale CMOS technologies,” Proc. IEEE, vol. 97, no. 10,
controlled latch sense amplifier and a static power-saving input buffer pp. 1687–1714, Oct. 2009.
for low-power architecture,” IEEE J. Solid-State Circuits, vol. 28, no. 4, [33] A. Papoulis, Probability, Random Variables, and Stochastic Processes,
pp. 523–527, Apr. 1993. 2nd ed. New York, NY, USA: McGraw-Hill, 1984.
[9] J. Kim, B. S. Leibowitz, J. Ren, and C. J. Madden, “Simulation and
analysis of random decision errors in clocked comparators,” IEEE Trans. Hao Xu (S’14–M’18) received the B.S. degree in
Circuits Syst. I, Reg. Papers, vol. 56, no. 8, pp. 1844–1857, Aug. 2009. mircoelectronics from Fudan University, Shanghai,
[10] L. G. Heller, D. P. Spampinato, and Y. L. Yao, “High sensitivity charge- China, in 2010, and the M.S. and Ph.D. degrees in
transfer sense amplifier,” IEEE J. Solid-State Circuits, vol. 11, no. 5, electrical engineering from the University of Califor-
pp. 596–601, Oct. 1976. nia at Los Angeles, in 2012 and 2018, respectively.
[11] K. Kotani, T. Shibata, and T. Ohmi, “CMOS charge-transfer preamplifier He has been with Broadcom, Inc., Irvine, since
for offset-fluctuation cancellation in low-power A/D converters,” IEEE 2017.
J. Solid-State Circuits, vol. 33, no. 5, pp. 762–769, May 1998. His research interests include mixed-signal, radio
[12] S. J. Mason, “Topological analysis of linear nonreciprocal networks,” frequency, and analog integrated circuit design.
Proc. IRE, vol. 45, no. 6, pp. 829–838, Jun. 1957.
[13] L. Chen, A. Sanyal, J. Ma, X. Tang, and N. Sun, “Comparator common-
mode variation effects analysis and its application in SAR ADCs,” Asad A. Abidi (S’75–M’80–SM’95–F’96) received
in Proc. IEEE Int. Symp. Circuits Syst., May 2016, pp. 2014–2017. the B.Sc. degree (Hons.) from Imperial College,
[14] P. Nuzzo, C. Nani, C. Armiento, A. Sangiovanni-Vincentelli, London, U.K., in 1976, and the M.S. and Ph.D.
J. Craninckx, and G. Van der Plas, “A 6-Bit 50-MS/s threshold config- degrees in electrical engineering from the University
uring SAR ADC in 90-nm digital CMOS,” IEEE Trans. Circuits Syst. of California at Berkeley, Berkeley, in 1978 and
I, Reg. Papers, vol. 59, no. 1, pp. 80–92, Jan. 2012. 1981, respectively. From 1981 to 1984, he was
[15] M. J. E. Lee, W. J. Dally, and P. Chiang, “Low-power area-efficient with Bell Laboratories, Murray Hill, NJ, USA, as a
high-speed I/O circuit techniques,” IEEE J. Solid-State Circuits, vol. 35, Member of the Technical Staff with the Advanced
no. 11, pp. 1591–1599, Nov. 2000. LSI Development Laboratory. Since 1985, he has
[16] B. Wicht, T. Nirschl, and D. Schmitt-Landsiedel, “Yield and speed been with the Electrical Engineering Department,
optimization of a latch-type voltage sense amplifier,” IEEE J. Solid-State University of California at Los Angeles, Los Ange-
Circuits, vol. 39, no. 7, pp. 1148–1158, Jul. 2004. les, where he is currently a Distinguished Chancellor’s Professor. He also
[17] L. Kull et al., “A 3.1 mW 8b 1.2 GS/s single-channel asynchronous holds the Abdus Salam Chair at LUMS, Lahore, Pakistan. His research
SAR ADC with alternate comparators for enhanced speed in 32 nm interests span fundamentals of circuit design, RF CMOS circuits, high-speed
digital SOI CMOS,” IEEE J. Solid-State Circuits, vol. 48, no. 12, analog circuits, and data conversion.
pp. 3049–3058, Dec. 2013. He has been elected Fellow of IEEE, a member of the U.S. National Acad-
[18] P. Nuzzo, F. De Bernardinis, P. Terreni, and G. Van der Plas, “Noise emy of Engineering, and a fellow of TWAS-the world academy of sciences.
analysis of regenerative comparators for reconfigurable ADC archi- He received an IEEE Millennium Medal, the 1988 TRW Award for Innovative
tectures,” IEEE Trans. Circuits Syst. I, Reg. Papers, vol. 55, no. 6, Teaching, the 1997 IEEE Donald G. Fink Award, the 2007 Lockheed-Martin
pp. 1441–1454, Jul. 2008. Award for Excellence in Teaching, and the 2008 IEEE Solid-State Circuit
[19] T. Sepke, P. Holloway, C. G. Sodini, and H.-S. Lee, “Noise analysis for Society’s Donald O. Pederson Award. He was named one of the top ten
comparator-based circuits,” IEEE Trans. Circuits Syst. I, Reg. Papers, contributors to the ISSCC in its first 50 and 60 years. From 1992 to 1995,
vol. 56, no. 3, pp. 541–553, Mar. 2009. he was an Editor-in-Chief of the IEEE J OURNAL OF S OLID -S TATE C IRCUITS .