Professional Documents
Culture Documents
Abstract—An ultra-low-power area-efficient 8-bit successive where Pin is the power drawn from the input, CT is the total
approximation register (SAR) analog-to-digital converter (ADC) capacitance of the array, VR is the reference voltage and Ts
is presented. To achieve ultra-low-power performance a DAC is the conversion period. The average power drawn form the
architecture is proposed that employs two rail-to-rail low-power
unity-gain buffers and only 4 minimum-size capacitors instead reference generator for charge redistribution is [7] :
of the conventional binary-weighted capacitor array. Thereby, CT VR2
power consumption and area are drastically reduced by virtue Pref = (2)
of lower switching activity and smaller size capacitor array. The 2Ts
proposed 8-bit SAR ADC is designed and simulated in a 0.13µm Accordingly, layout area and power consumption grow ex-
CMOS process. Simulation results show that for a 2.4 kHz
(12.4 kHz) input signal while sampling at 25 kHz, the ADC
ponentially (CT = 2n C0 ) with the number of bits (n) which
achieves an ENOB of 7.9 (7.8), consumes 290 nW (350 nW) form compromises the advantage of SAR ADCs in terms of area
a 0.8 V analog supply and a 0.6 V digital supply, and achieves and power, especially as the number of bits increases.
a FoM of 48 fJ/conversion-step (62 fJ/conversion-step). Extensive studies have been carried out on energy-efficient
I. I NTRODUCTION DAC structures [1-8]. [2] proposes the capacitor splitting
scheme which reduces the average switching energy of the
Emerging mixed-signal applications such as radio-frequency array by splitting the most significant bit (MSB) capacitor
identification (RFID) systems, wireless sensor networks, into n − 1 binary scaled sub-capacitors (where n is the
portable biometric devices and energy scavenging systems number of bits). Consequently, for down transitions [2], the
demand a high power efficiency to secure a long autonomous charge on the MSB capacitor will be restored for further
operation. Analog-to-digital converters (ADCs) are among conversions. Although the proposed technique significantly
critical components that are used in most of such systems saves the switching energy for small output codes, it fails to
and usually consume a fair share of the overall power budget. perform equally well for higher output codes where less down
Among different ADC architectures, successive approximation transitions occur. Moreover, this approach is still dependant
register (SAR) ADCs [1-6] are the most popular candidate for on a bulky DAC array and requires twice as many switches
the aforementioned applications which typically require high- as the conventional scheme.
resolution, low-to-medium speed and low power. In another structure, [8] reduces the average switching
Conventional SAR ADCs use the standard binary search energy of the capacitive array by separating the decoding
technique [1-6]. Generally, in a SAR ADC there are two of MSB and least significant bit (LSB) through using two
major contributors to power consumption: comparator and different capacitor arrays with unequal size. However, the low-
charging/discharging of the DAC array. However, comparator’s power consumption is achieved at the cost of two additional
power consumption is usually not a matter of concern unless clock cycles which poses energy overhead due to additional
a very high-speed and high resolution is required. (Yet, a comparison steps required. Moreover, the layout occupies a
comparator architecture compatible with the proposed digital- higher area than that of the traditional architecture due to MSB
to-analog converter (DAC) structure is briefly discussed in the DAC, memory storage with logic and the extra comparator.
paper.) This paper proposes an area-efficient DAC architecture that
Typical SAR structures use bulky DAC capacitive arrays both reduces the power consumption and the layout area. The
which are usually the area and power bottleneck of the design. proposed DAC architecture consists of four minimum-size
From the power dissipation perspective, in order to sample a capacitors and two low-power buffers for which the binary
full-scale sine wave at the Nyquist frequency, a traditional search algorithm is carried out at no extra cost in terms of
capacitive array on average consumes [7]: clock cycles and comparison steps. Also, the use of four
CT VR2 minimum-sized capacitor DAC relaxes capacitor mismatch
Pin = (1)
Ts and parasitic requirements. To further save area and power
consumption, the two low-power buffers are dually used as
This research is funded in part by the Natural Sciences and Engineering the comparator preamplifier blocks in an open-loop fashion.
Research Council of Canada (NSERC) and the Institute for Computing,
Information and Cognitive Systems (ICICS) at UBC. CAD support and access Section II presents the DAC architecture while Section III
to technology is facilitated by CMC Microsystems. describes the proposed comparator. In Section IV, the sample-
Authorized licensed use limited to: INDIAN INSTITUTE OF TECHNOLOGY KHARAGPUR. Downloaded on February 11,2021 at 15:34:32 UTC from IEEE Xplore. Restrictions apply.
VREF VREF
Ctop1 Ctop2
S1 S3 S5 S7
VTOP1 VTOP2
St1 VMID1 B1 St2
VMID2
Sb1 B2 Sb2
VBOT1 VBOT2
14
Authorized licensed use limited to: INDIAN INSTITUTE OF TECHNOLOGY KHARAGPUR. Downloaded on February 11,2021 at 15:34:32 UTC from IEEE Xplore. Restrictions apply.
CLKcomp VREF VREF
OUT+
CLKcomp Mp1 CLK-latch
VIN Mp2
OUT-
CLKcomp1 OUT+
B1 OUT OUT+
VMID1
CLK-pre
CLKcomp2 CLKcomp2 CLKcomp
VMID2 VMID2
CLK-latch SL OUT
CLKcomp1 CLKcomp1,2
VMID1
OUT- CLKcomp CLK-pre
B2 VIN CLK-latch
OUT-
CLK- pre Mn1 CLK-latch Mn2
CLK- off
CLKcomp
Fig. 3. Proposed comparator architecture and timing diagram. Fig. 4. Low-power latch architecture.
VREF
the conventional SAR architectures. As the algorithm suggests,
during a conversion cycle, each of the two buffers contribute
in the charge recycling process for at most n times. So, the IB OUT
total charge drawn from VREF during the conversion cycle is: VIN- VIN+ CC
15
Authorized licensed use limited to: INDIAN INSTITUTE OF TECHNOLOGY KHARAGPUR. Downloaded on February 11,2021 at 15:34:32 UTC from IEEE Xplore. Restrictions apply.
VREF
CLKH CLKH
Crefresh
CLKH CLKH
Cboost Sampling
Period Fig. 7. Simulated DNL and INL at 25kS/s.
CLKSAMPLE MH2 MH1 CLKSAMPLE
VIN
G
V
Msample
Mdummy
Csample
CLKSAMPLE
16
Authorized licensed use limited to: INDIAN INSTITUTE OF TECHNOLOGY KHARAGPUR. Downloaded on February 11,2021 at 15:34:32 UTC from IEEE Xplore. Restrictions apply.