54 IEEE JOURNAL OF SOLID-STATE CIRCUITS, VOL. 38, NO. 1, JANUARY 2003
A New Successive Approximation Architecture for
Low-Power Low-Cost CMOS A/D Converter
Chi-Sheng Lin and Bin-Da Liu, Senior Member, IEEE
Abstract—A new 6-bit 250 MS/s analog-to-digital converter
(ADC) is proposed for low-power low-cost CMOS integrated
systems. This design is based on an improved successive ap-
proximation ADC with a mixed-mode subtracter that minimizes
the overall power consumption and system complexity. The
experimental results indicate that this ADC works up to 250 MS/s
with power consumption less than 30 mW at 3.3 V. Moreover, the
operating voltage is scaled down to 0.8 V using a slight adjustment.
The ADC occupies only 0.1 mm
2
with the TSMC 0.35- m single
ploy quadruple metal (SPQM) CMOS technology. This design is
suitable for standard CMOS technology with low-power low-cost
VLSI implementation. It is well applied when embedded into
system-on-chip (SoC) circuit designs.
Index Terms—ADC, low voltage, parallel-like, successive
approximation.
I. INTRODUCTION
H
IGH performance CMOS analog-to-digital converters
(ADCs) are key components in mixed-signal integrated
circuits. Recent ADC applications are used increasingly in
digital data reading fields, such as hard disk drives, digital
videodiscs and local-area networks [1], [2]. High sampling
speed is required in all of these applications, resolution as low
as 6-bit, however, is sufficient.
Among these low-resolution high-speed CMOS ADC design
techniques, the flash architecture is one of the best solutions
for high-speed low-latency operations [3]. However, with the
number of bits
, the flash architecture requires com-
parators and the power, area and input capacitance are propor-
tional to
. Folding CMOS ADC is another good architec-
ture for high-speed operations and simple hardware design. This
is a technique for reducing the number of comparators used in
the flash architecture [4]. However, their performance degrades
with process variation [3].
In portable system designs, a low-voltage operation plays one
of the key design factors. The voltage limitations of this tech-
nology dictate that the ADC operates at the same low voltage
as the digital circuitry. In the literature, many low-voltage
CMOS ADC integrated circuits have been proposed [5], [6].
However, most of them are more complex circuit designs or
require higher power consumption. Consequently, they are not
suitable for portable system designs.
Manuscript received April 3, 2002; revised August 5, 2002. This work was
supported in part by the Chip Implementation Center, National Science Council
and by the Program of Promoting Academic Excellence, Ministry of Education,
under Grant EX-91-E-FA09-5-4, Taiwan, R.O.C.
The authors are with the Department of Electrical Engineering, Na-
tional Cheng Kung University, Tainan 70101, Taiwan, R.O.C. (e-mail:
bdliu@cad.ee.ncku.edu.tw).
Digital Object Identifier 10.1109/JSSC.2002.806257
Fig. 1. Circuit diagram of a general successive approximation ADC.
In this paper, a novel circuit for low-power low-cost 6-bit
CMOS ADC is presented. A basic successive approximation
ADC (SA-ADC) will first be introduced and followed by an im-
proved successive approximation ADC (ISA-ADC) with stan-
dard CMOS technology. Based on the ISA-ADC architecture, a
parallel-like ISA-ADC architecture for high-speed low-resolu-
tion applications is developed.In addition, the adjustable feature
of the proposed circuit for scaling down the operating voltage
will be shown. The proposed converter has a simple hardware
design and low-accuracy comparator and therefore, is suitable
for low-power low-cost standard CMOS technology VLSI im-
plementation.
II. B
ASIC ARCHITECTURES OF SUCCESSIVE
APPROXIMATION ADC
The architecture of a general SA-ADC usually consists of
a rail-to-rail analog comparator, a digital-to-analog converter
(DAC)andasuccessiveapproximationregister (SAR), as shown
in Fig. 1 [7]. In general, the SAR is designed into the digital
circuitry. The DAC required in the ADC is designed based on
a R-2R ladder. Thus, both of them are suitable for the stan-
dard CMOS technology VLSI implementation. However, real-
izing a high-speed, high-accuracy and rail-to-rail MOS com-
parator concurrently remains a problem because of MOS device
mismatches [8], [9] and threshold voltage limitations [10]. The
input signals of this comparator is expressed by
(1)
where both
and are the comparator inputs and denotes
the comparator operation. This high-speed, offset compensated
and rail-to-rail comparator plays one of the key components for
SA-ADC designs.
To solve this problem, another SA-ADC architecture was de-
veloped to simplify the comparator requirement, as shown in
Fig. 2 [11]. This architecture employs an additional analog sub-
tracter to generate the differential value between the
signal
and the
signal, this way a comparator based on this archi-
tecture is applicable to compare the differential voltage level
0018-9200/03$17.00 © 2003 IEEE