IEEE TRANSACTIONS ON CIRCUITS AND SYSTEMS—II: EXPRESS BRIEFS, VOL. 61, NO. 6, JUNE 2014 403
Digital Noise-Coupling Technique for Delta–Sigma
Modulators With Segmented Quantization
Lin He, Member, IEEE, Yuncheng Zhang, Fang Long, Fengcheng Mei, Mingyuan Yu,
Fujiang Lin, Senior Member, IEEE , Libin Yao, Senior Member, IEEE, and Xicheng Jiang, Fellow, IEEE
Abstract—A digital noise-coupling technique for delta–sigma
modulators with a high-resolution quantizer is presented. The
proposed technique divides the outputs of the quantizer into an
MSB segment and an LSB segment. The MSBs are directly fed to
the modulator input, whereas the LSBs are used as the quantiza-
tion noise for noise coupling. After the digital postprocessing, the
MSBs and the LSBs are then recombined. As a result, it preserves
the benefit of increased quantization levels without suffering from
the exponential growth of the dynamic element matching logic.
The extra noise shaping introduced by the LSB feedback further
reduces the quantization noise leakage caused by the mismatch
between the analog and digital noise transfer functions, which al-
lows a more relaxed operational amplifier design. This technique is
simple to implement and well suited for low-power and wideband
applications.
Index Terms—Delta–sigma (ΔΣ), dynamic element match-
ing (DEM), digital noise coupling, large quantization level,
Leslie–Singh, low-power, noise cancelation logic, noise coupling,
segmented quantization, wideband.
I. INTRODUCTION
D
ELTA–sigma (ΔΣ) modulation is a typical high-
resolution analog-to-digital conversion (ADC) architec-
ture that finds wide applications in communication systems.
There is an increasing demand for low-power ΔΣ ADCs to
extend the battery lifetime. Power saving can be done either at
the system level, such as full feedforward [1], noise coupling
[2], operational amplifier (op-amp) sharing [3], continuous-
time feedback [4]–[6], and double sampling [7], or at the
circuit level, such as the inverter-based structure [8]. Among
the various low-power circuit design techniques, the most
straightforward one is to increase the resolution of the multi-
bit quantizer, as it reduces the quantization noise as well
as the output signal swings of the integrators, so that the
slew rate and the gain requirement of the op-amps are re-
Manuscript received February 9, 2014; revised March 24, 2014; accepted
April 10, 2014. Date of publication April 24, 2014; date of current version
June 13, 2014. This work was supported in part by the National Natural Science
Foundation of China under Project 61204033 and in part by the Natural Science
Foundation of Jiangsu Province under Project BK2012214. This brief was
recommended by Associate Editor L. H. Corporales.
L. He, Y. Zhang, F. Long, F. Mei, M. Yu, and F. Lin are with the Micro-/
Nano-Electronic System Integration Center (MESIC), University of Science
and Technology of China, Hefei 230026, China (e-mail: helin77@ustc.edu.cn;
zhangyc23@gmail.com; longfang124@gmail.com; fchmei@mail.ustc.edu.cn;
ymyxinde@126.com; linfj@ustc.edu.cn).
L. Yao is with Kunming Institute of Physics, Kunming 650223, China
(e-mail: libin.yao@gmail.com).
X. Jiang is with Broadcom Corporation, Irvine, CA 92617 USA (e-mail:
jerry.jiang@gmail.com).
Color versions of one or more of the figures in this brief are available online
at http://ieeexplore.ieee.org.
Digital Object Identifier 10.1109/TCSII.2014.2319994
laxed, which is particularly useful for scaled CMOS technol-
ogy [9]. However, increasing the resolution of a conventional
Flash quantizer requires an exponential growth of compo-
nents and power. For this reason, the resolution of most of
the Flash quantizers in a ΔΣ modulator is usually limited
to 4 bits. This problem can be partially solved by the intro-
duction of different types of quantizers, such as a two-step
quantizer [10], a pipelined quantizer [11], a synthetic aperture
radar (SAR) quantizer [12], [13], or a nonuniform quantizer
[14], [15].
However, the dynamic element matching (DEM) logic,
which dynamically shuffles the digital-to-analog conversion
(DAC) elements to convert their mismatch into a shaped noise,
also suffers from the exponential growth problem. Digital ΔΣ
modulation [10], [12], [16], [17] can be used to reduce the num-
ber of feedback levels. However, this digital modulation tech-
nique introduces not only an extra power but also a logic delay,
which will limit the operation frequency of the whole system.
Reference [11] directly fed back the digital outputs of the first
pipelined stage to the modulator input to reduce the complexity
of the feedback DAC and introduced many feedback paths for
higher order noise shaping. Such an approach is complicated
and occupies a large chip area. Reference [13] adopted the
well-known Leslie–Singh architecture [18]–[20] and divided
the outputs of the quantizer into an MSB segment and an LSB
segment. The MSBs were directly fed to the modulator input,
and a digital cancelation logic was used to combine the MSBs
and the LSBs. This architecture ideally is equal to a modulator
with a high-resolution quantizer but, in practice, suffers from
the quantization noise leakage problem caused by the mis-
match between the analog and digital noise transfer functions,
which caused a measured signal-to-noise-plus-distortion ratio
(SNDR) degradation of more than 10 dB [13].
In this brief, we propose a general digital noise-coupling
technique for ΔΣ modulators with a high-resolution quantizer.
The proposed technique allows reduced feedback levels to
simplify the DEM logic. It preserves the simplicity of the
Leslie–Singh architecture to avoid the logic delay problem in
the digital modulation technique, but is much less sensitive
to the mismatch between the analog and digital noise transfer
functions. This brief is organized as follows. Section II briefly
reviews the existing noise-coupling technique, the Leslie–Singh
architecture, and introduces the proposed digital noise-coupling
architecture. Section III analyzes the impact of circuit im-
perfections on the performance of the proposed architecture.
Behavioral simulations are performed to verify the analysis
using a second-order low-pass ΔΣ modulator as an example.
Section IV discusses the implementation issues. The conclusion
is provided in Section V.
1549-7747 © 2014
IEEE. Personal use is permitted, but republication/redistribution requires IEEE permission.
See http://www.ieee.org/publications_standards/publications/rights/index.html for more information.