sensors Article

A Reconfigurable Readout Integrated Circuit for Heterogeneous Display-Based Multi-Sensor Systems Kyeonghwan Park, Seung Mok Kim, Won-Jin Eom and Jae Joon Kim * School of Electrical and Computer Engineering, Ulsan National Institute of Science and Technology, 44919 Ulsan, Korea; [email protected] (K.P.); [email protected] (S.M.K.); [email protected] (W.-J.E.) * Correspondence: [email protected]; Tel.: +82-52-217-2126 Academic Editor: Vittorio M. N. Passaro Received: 6 February 2017; Accepted: 31 March 2017; Published: 3 April 2017

Abstract: This paper presents a reconfigurable multi-sensor interface and its readout integrated circuit (ROIC) for display-based multi-sensor systems, which builds up multi-sensor functions by utilizing touch screen panels. In addition to inherent touch detection, physiological and environmental sensor interfaces are incorporated. The reconfigurable feature is effectively implemented by proposing two basis readout topologies of amplifier-based and oscillator-based circuits. For noise-immune design against various noises from inherent human-touch operations, an alternate-sampling error-correction scheme is proposed and integrated inside the ROIC, achieving a 12-bit resolution of successive approximation register (SAR) of analog-to-digital conversion without additional calibrations. A ROIC prototype that includes the whole proposed functions and data converters was fabricated in a 0.18 µm complementary metal oxide semiconductor (CMOS) process, and its feasibility was experimentally verified to support multiple heterogeneous sensing functions of touch, electrocardiogram, body impedance, and environmental sensors. Keywords: heterogeneous multi-sensor functions; alternate sampling ADC; reconfigurable readout; touch sensor; environment sensors; electrocardiogram; body impedance

1. Introduction Display technology innovations have been focused primarily on their resolution and size, but another remarkable trend is that they are increasingly embedding more different functions such as touch sensors [1]. Among various touch sensors, the mutual-capacitance touch-sensing method has been popular, especially in smartphone applications, and its derivative studies have enhanced touch sensitivity and immunity against noise and interference [2,3]. Recent touch screen panels (TSPs) have become integrated inside display panels. Moreover, there have been recent efforts to implement the fingerprint recognition function on the TSP [4,5]. In flexible or wearable application fields, there have been device-based trials to embed different kinds of signal acquisitions, including bio-signal, respiration and human movements, through various types of sensors [6,7]. Considering these different research trends, this work tries to propose a kind of display-based multi-sensor interface that can embed various sensors from heterogeneous fields. That is, in addition to the inherent touch function of the TSP, two different heterogeneous interfaces of physiological and environmental sensors are proposed. For compact implementation that supports various heterogeneous sensors effectively, its common readout integrated circuit (ROIC) is required to have a reconfigurable interface structure. Additionally, it needs to provide proper remedies against noises and motion that its inherent TSP-based operation accompanies. Accordingly, a TSP-based multi-sensor interface is proposed to support various heterogeneous sensors including electrocardiogram (ECG), body impedance (BI) [8,9], and also environmental sensors that are mostly resistive or capacitive [10,11]. In order to accommodate these heterogeneous sensors, Sensors 2017, 17, 759; doi:10.3390/s17040759

www.mdpi.com/journal/sensors

Sensors 2017, 17, 759

2 of 13

the ROIC is designed to have a reconfigurable structure that consists of an amplifier-based signal path and an oscillator-based signal path. The amplifier-based path provides two sensor interfaces for touch and ECG, and the oscillator-based path gives reconfigurable means to support BI and environmental sensors. This classification has been done to optimize their required performance and processing cost of power and area. In the case of BI detection, while there have been several sensing methods, such as multi-frequency analysis, capacitive measurement, and current injection [12–14], this work selected the oscillator-based method to support other kinds of sensors. The oscillator-based path can provide direct digital conversion from resistive or capacitive sensor variations, called X-to-digital converters (XDCs), which substitutes conventional analog-to-digital converters (ADCs). Therefore, depending on each sensor’s operating path, the ADC or the XDC are selectively utilized for its digital conversion. For the ADC implementation, successive approximation register (SAR) types have been used widely in low-power applications due to their excellence in power efficiency compared with other ADC architectures. Conventional SAR ADCs consist of capacitive digital-to-analog converters (C-DACs), a dynamic comparator, and control logics, which do not require static power consumption. In order to improve noise immunity and signal-to-noise ratio (SNR), the SAR ADC is reformed to include a proposed scheme of alternate sampling (AS) and error correction. The remainder of this paper is organized as follows. In Section 2, we present the proposed display-based multi-sensor interface architecture and describe the alternate-sampling error-correction scheme. Section 3 explains circuit implementation of the overall function following each sensor-signal path. Section 4 shows experiment results of its prototype design including the reconfigurable ROIC and AS-SAR ADC. Finally, the conclusion is given in Section 5. 2. Display-Based Multi-Sensor Interface 2.1. Heterogeneous Architecture Figure 1 presents a conceptual diagram of the proposed reconfigurable readout integrated circuit (ROIC) for heterogeneous sensors, also including a system prototype on a TSP that has 32 transmit (TX) lines and 8 receive (RX) lines. While the main role of the previous TSP interface was to detect various touch events, the proposed architecture utilizes the TSP itself as a kind of multi-purpose sensing channel, where different sensing functions are achieved depending on its readout circuit configuration. For effective implementation of the display-based multi sensor system, whole sensor readout topologies are consolidated into the following two basis readout circuits: amplifier-based and oscillator-based. That is, since most TSP readouts are implemented with capacitive amplifiers and weak ECG signals need to be amplified, they are merged together into a single amplifier-based readout circuit, utilizing the TSP cells as capacitive electrodes for the ECG operation. Remaining functions of BI and environmental sensors are unified into the oscillator-based readout circuit, considering that the one requires frequency generation and the other can be detected by using oscillators. Depending on their applications, these heterogeneous sensor functions are classified into two types: stretchable and bio-signal user interfaces (UIs). The stretchable UI can detect capacitive and resistive variations, and it can be utilized to support various user interfaces such as touch, stretching, or bending, especially in flexible display applications. Moreover, many environmental sensors have capacitive or resistive sensor characteristics [15]. Excluding the touch interface that uses the amplifier-based readout, these resistive and capacitive sensing interfaces are provided by making the oscillator-based readout frequency respond to variations of resistance or capacitance. Then, its digital conversion is done by utilizing a counter, and this operation is called the XDC, which does not require additional ADCs. The bio-signal UI is designed to provide physiological signals such as BI and ECG. The ECG interface is supported by the amplifier-based readout circuit, utilizing TSP cells as capacitive electrodes. For better capacitive signal coupling, the TSP cells are tied together to formulate bigger capacitance. The BI interface is provided by the oscillator-based readout circuit. Since the BI requires multiple frequency sweeps, the oscillator frequency is changed by programming internal

Sensors 2017, 17, 759

Sensors 2017, 17, 759

3 of 13

3 of 14

requires multiple frequency sweeps, the oscillator frequency is changed by programming internal

control bits. For noise-immune design, the proposed ROIC includes an alternate-sampling successive control bits. For noise-immune design, the proposed ROIC includes an alternate-sampling successive approximation register (AS-SAR) error-correctioncapability. capability. Then, is used for the approximation register (AS-SAR)ADC ADCwhich which has has error-correction Then, it isitused for the digital conversion of the bio-signal interface,while whilethe the stretchable excluding digital conversion of the bio-signalUI UIand andthe the touch touch interface, stretchable UI UI excluding the the touch is supported by the XDC. touch is supported by the XDC.

Figure 1. Display-based multi-sensor interface and reconfigurable readout integrated circuit.

Figure 1. Display-based multi-sensor interface and reconfigurable readout integrated circuit.

2.2. Alternate-Sampling Error-Correction Scheme

2.2. Alternate-Sampling Error-Correction Scheme

Many studies have presented ways for achieving better resolution while maintaining the lowManycharacteristic studies have presented ways In fororder achieving better for resolution maintaining power of SAR ADC [16–19]. to compensate mismatchwhile and non-linearity in the low-power of SAR ordertechnologies to compensate for dithering mismatch C-DACs, characteristic which mainly limit overallADC ADC [16–19]. resolution,Invarious including [16] and non-linearity in C-DACs, which mainly limit ADC resolution, various technologies including and digital-domain calibration [17] have beenoverall used. Hybrid architectures with integrating ADCs [18] or oversampling ADCs [19] have also been adopted to partially utilize high-resolution technologies. dithering [16] and digital-domain calibration [17] have been used. Hybrid architectures with integrating However, previous works complex circuits supplementary conversion ADCs [18] or these oversampling ADCsrequire [19] have alsoadditional been adopted toand partially utilize high-resolution [20,21], and an alternate-sampling (AS) SARrequire ADC structure proposed circuits to relieve this difficulty. technologies. However, these previous works complexisadditional and supplementary Figure 2a describes the proposed AS scheme in SAR ADCs and how to provide error-correction conversion [20,21], and an alternate-sampling (AS) SAR ADC structure is proposed to relieve capability. The ADC circuit is designed to be fully differential, but its input path from various sensor this difficulty. Figure 2a describes the proposed AS scheme in SAR ADCs and how to provide outputs is single-ended. Therefore, the first role of the AS scheme is to convert the single-ended error-correction capability. The ADC circuit is designed to be fully differential, but its input path from sensor input into its corresponding differential signal which can improve common-mode noise various sensor outputs is single-ended. Therefore, the first role of the AS scheme is to convert the immunity considerably. The second role is that the converted differential data can be used for error single-ended input into its corresponding which improve common-mode correctionsensor by comparing it with an interpolated differential values from signal adjacent data. can In this way, the proposed noise The second role that theand converted datacorrection. can be used ASimmunity scheme canconsiderably. provide better common-mode noiseisimmunity also a newdifferential function of error

for error correction by comparing it with an interpolated values from adjacent data. In this way, the proposed AS scheme can provide better common-mode noise immunity and also a new function of error correction.

Sensors 2017, 17, 759

4 of 13

Sensors 2017, 17, 759

4 of 14

(a)

(b)

Figure 2. (a) Alternate-sampling error-correction scheme and (b) its circuit implementation.

Figure 2. (a) Alternate-sampling error-correction scheme and (b) its circuit implementation.

The alternate sampling method is conceptually based on correlated double sampling (CDS) The alternate method is conceptually on correlated double sampling (CDS) technology [22]. sampling As shown in Figure 2b, it operates likebased conventional differential SAR ADCs, except that polarity differential input 2b, is alternately by the end-of-conversion (EOC) signal. Two that technology [22]. of Asa shown in Figure it operatesinverted like conventional differential SAR ADCs, except inputofpaths are connected or crossly depending onend-of-conversion their control signal(EOC) switchsignal. (SW), and polarity a differential inputdirectly is alternately inverted by the Twothe input input sampling polarity is inverted whenever the EOC signal is activated. That is, two differential paths are connected directly or crossly depending on their control signal switch (SW), and the input inputs polarity are then issampled onto C-DACsthe in phase or anti-phase, and then are differential compared ininputs a sampling inverted whenever EOC signal is activated. Thatthey is, two dynamic comparator whose output adjusts the next-cycle value of C-DACs. In this way, the are then sampled onto C-DACs in phase or anti-phase, and then they are compared in a dynamic successive approximation procedure of N cycles is performed to make the residue smaller than half comparator whose output adjusts the next-cycle value of C-DACs. In this way, the successive of the least significant bit (LSB) value. After this N-cycle SAR conversion, N-bit ADC outputs are approximation procedure of N cycles is performed to make the residue smaller than half of the obtained and the EOC flag is activated for one cycle. The rising edge of the EOC inverts the polarity leastofsignificant bit (LSB) value. After this N-cycle SAR conversion, N-bit ADC outputs are obtained the next differential input sampling, which is again reversed at the end of the next conversion and period. the EOC flag is activated for one cycle. The rising edge of the EOC inverts the polarity of the next differential input sampling, which is again reversed atperiodic the endphase of thealternation next conversion period.input In this way, the proposed AS-SAR ADC performs in differential In this way, theconventional proposed AS-SAR ADC performs periodic phase are alternation in First, differential sampling during SAR conversion. Four detailed procedures as follows. ininput sampling during conventional SAR are conversion. detailed procedures are as data, follows. phase and anti-phase differential inputs alternately Four sampled and converted to digital in and incomplete differential digital signals. alternately vacant data in sampled First,resulting in-phase anti-phase differential inputs are Second, alternately sampled and converted to digital differential waveforms are interpolated from adjacent sample points. Then, the interpolated values data, resulting in incomplete differential digital signals. Second, alternately vacant data in sampled from in-phase sampled can be converted into its corresponding anti-phase values which canvalues be differential waveforms aredata interpolated from adjacent sample points. Then, the interpolated compared with anti-phase sampled data. Through these comparisons between the sample data and from in-phase sampled data can be converted into its corresponding anti-phase values which can be the interpolated values, instant errors in each sample can be detected as:

compared with anti-phase sampled data. Through these comparisons between the sample data and (𝑖) =can (𝑖) + 𝑒𝑖𝑛𝑠𝑡𝑎𝑛𝑡 𝐷𝑆𝑎𝑚𝑝𝑙𝑒𝐴 𝐷𝑠𝑎𝑚𝑝𝑙𝑒𝐵 𝐷max (1) the interpolated values, instant errors(𝑖)in+each sample be𝑐𝑜𝑑𝑒 detected as: ,

where 𝐷𝑆𝑎𝑚𝑝𝑙𝑒𝐴 (𝑖) and 𝐷𝑠𝑎𝑚𝑝𝑙𝑒𝐵 (𝑖) are 𝑖 th differentially-sampled data, and instant errors of DSampleA (i ) + DsampleB (i ) = Dmaxcode ((𝑖). i ) + einstant , (1) 𝑒𝑖𝑛𝑠𝑡𝑎𝑛𝑡 occur along to their peak-to-peak amplitude of 𝐷max 𝑐𝑜𝑑𝑒 Without errors, sum of in-phase and anti-phase sampled data becomes equal to 𝐷max 𝑐𝑜𝑑𝑒 (𝑖). The ADC has an average power of where DSampleA (i ) and D2sampleB (i ) are ith differentially-sampled data, and instant errors of einstant quantization noise VLSB /12 and the magnitude of acceptable error is smaller than 0.5 LSB [23]. Then, occur peak-to-peak amplitude of Dmaxcode . Without errors, sumitsofcounterpart in-phase and if along 𝑒𝑖𝑛𝑠𝑡𝑎𝑛𝑡toistheir larger than 0.5 LSB, the erroneous sample(i )can be corrected with anti-phase sampled data becomes equal to D i . The ADC has an average power of quantization ( ) maxcode interpolated value that is estimated from its adjacent anti-phase samples. The proposed scheme can 2 noise VLSB /12consecutive and the magnitude of acceptable error is smaller than 0.5 LSB [23]. Then, if einstant also correct errors thanks to the inherent splitting operation of the alternate sampling. If is larger than 0.5two LSB, the erroneous sample can be corrected itsthem counterpart interpolated value that there are consecutive errors, the proposed work can with correct by alternating two sampled values and comparing themanti-phase with their samples. corresponding interpolated values. Asalso a final procedure, a is estimated from its adjacent The proposed scheme can correct consecutive signal doubling achievedoperation by takingof the difference two anti-phased is errors thanks to the function inherentissplitting the alternateofsampling. If theresamples, are two which consecutive

errors, the proposed work can correct them by alternating two sampled values and comparing them with their corresponding interpolated values. As a final procedure, a signal doubling function is

Sensors 2017, 17, 759

Sensors 2017, 17, 759

5 of 13

5 of 14

achieved by taking the difference of two anti-phased samples, which is inherently performed in differential SAR ADC conversion. While differential signaling provides common-mode inherently performed in differential SARconventional ADC conversion. While conventional differential signaling noise rejection and effectivenoise signal doubling ×), the signal proposed scheme additional noise provides common-mode rejection and(2effective doubling (2×),provides the proposed scheme rejection and quadruple increments of effective signal strength (4 × ), resulting in meaningful provides additional noise rejection and quadruple increments of effective signal strength (4×),SNR improvement. the improvement. effects of offset cancellation low-frequency rejectionand in the resulting in Considering meaningful SNR Considering theand effects of offset cancellation low-CDS frequency rejection in the CDS technology, this proposed AS-SAR supposed to reduce input technology, this proposed AS-SAR ADC is supposed to reduce inputADC offsetisand its low-frequency noises. offset and its low-frequency noises.

3. ROIC Implementation 3. ROIC Implementation

3.1. Design of Oscillator-Based Readout Circuit 3.1. Design of Oscillator-Based Readout Circuit

The proposed display-based multi-sensor readout implementations are consolidated into two The proposed display-based multi-sensor readout are consolidated into two base circuits: oscillator-based and amplifier-based. First, implementations the oscillation-based readout interface that is base circuits: oscillator-based and amplifier-based. First, the oscillation-based readout interface that as shown Figure 3 is selectively configured to utilize a resistance- or capacitance-controlled oscillator is shown Figure 3 is selectively configured to utilize a resistance- or capacitance-controlled oscillator the XDC for resistive/capacitive sensors or the signal source for the BI measurement. The oscillator as the XDC for resistive/capacitive sensors or the signal source for the BI measurement. The oscillator consists of a charge pump and a delay cell, and its charging/discharging speed can be controlled consists of a charge pump and a delay cell, and its charging/discharging speed can be controlled by by load capacitance or by its current-mirror reference current. After some logic delay (τ), feedback load capacitance or by its current-mirror reference current. After some logic delay (τ), feedback connection of digitally controlled constitutesa aking kingofof ring oscillator connection of digitally controlledoscillator oscillator (DCO) (DCO) constitutes ring oscillator [24],[24], andand it it converts the charging/discharging speed movement into the frequency displacement which can converts the charging/discharging speed movement into the frequency displacement which can be be converted to digital values processagainst againsta afixed-frequency fixed-frequency external clock converted to digital valuesthrough throughthe thecounting counting process external clock (CLK_REF). In this way,way, the proposed oscillator-based interface can can work as the and also readout (CLK_REF). In this the proposed oscillator-based interface work as XDC the XDC and also readoutorcapacitive resistive sensor variations, without requiring additional capacitive resistiveor sensor variations, without requiring additional ADCs.ADCs. On the other hand, measurementis is performed performed by body impedance at multiple On the other hand, thethe BIBImeasurement bymeasuring measuring body impedance at multiple frequencies that are available from the same oscillator as in the XDC. Like multi-frequency frequencies that are available from the same oscillator as in the XDC. Like multi-frequency bioelectrical bioelectrical impedance analysis [12], low frequency 1 k are to 500 kHz for the BI impedance analysis [12], low frequency ranges of 1 kranges to 500of kHz used forare theused BI measurement. measurement. The human body consists of resistance and capacitance from the cell membranes, The human body consists of resistance and capacitance from the cell membranes, and the bodyand model the body model is composed of their parallel combination. The lower part of Figure 3 presents the BI is composed of their parallel combination. The lower part of Figure 3 presents the BI measurement measurement configuration that utilizes TSP cells as capacitive electrodes. Since each TSP cell has configuration that utilizes TSP cells as capacitive electrodes. Since each TSP cell has small capacitance small capacitance and contact area, whole TSP cell’s TX/RX electrodes are merged together to and maximize contact area, whole capacitance TSP cell’s TX/RX electrodes merged together to maximize mutual the mutual and possible contactare area. The electrical characteristic thatthe passes capacitance and possible contact area. The electrical characteristic that passes through the human through the human body is measured by a peak detector and then converted to digital value by thebody is measured by a peak detector and then converted to digital value by the AS-SAR ADC. AS-SAR ADC.

Figure 3. Oscillator-basedmulti-sensor multi-sensor interface interface configuration readout circuit. Figure 3. Oscillator-based configurationand and readout circuit.

Sensors 2017, 17, 759

6 of 13

Sensors 2017, 17, 759

6 of 14

3.2. Design of Amplifier-Based Readout Circuit 3.2. Design of Amplifier-Based Readout Circuit

Figure 4 presents the amplifier-based multi-sensor configuration and its reconfigurable readout presents multi-sensor and its reconfigurable circuit forFigure touch4and ECG.the Foramplifier-based the touch operation, touch configuration events are converted to capacitancereadout variations circuit for touch and ECG. For the touch operation, touch events are converted to capacitance through the TSP, and these capacitive changes are sensed by adjusting the capacitive amplifier’s gain variations through the TSP, and these capacitive changes are sensed by adjusting the capacitive on mutually coupled signals from the TX signal. For compact design of the ROIC, every TSP cell is amplifier’s gain on mutually coupled signals from the TX signal. For compact design of the ROIC, multiplexed and supported by a single capacitive amplifier. For high-precision touch detection, it is every TSP cell is multiplexed and supported by a single capacitive amplifier. For high-precision touch followed by a low-pass filter (LPF) to remove various environmental noises. A band-reject filter (BRF) detection, it is followed by a low-pass filter (LPF) to remove various environmental noises. A bandis selectively toisremove power-supply noise of 50 or 60 Hz [25]. reject filterused (BRF) selectively used to remove power-supply noise of 50 or 60 Hz [25].

(a)

(b) Figure 4. (a) Amplifier-based multi-sensor interface configuration, and (b) readout circuits.

Figure 4. (a) Amplifier-based multi-sensor interface configuration; and (b) readout circuits.

For the ECG detection, the TSP is configured to provide two capacitive electrodes by dividing Forpanel the ECG detection, the TSP configured to provide two capacitive dividing the by half and merging the is two resulting half-panel’s TX and RX lines,electrodes as shown by in the lower the panel byof half and4.merging the two resulting half-panel’s TXsignal and [26], RX lines, as bandwidth shown in the lower part part Figure Considering 300 Hz bandwidth of the ECG the LPF is adjusted through After the BRF additional amplification, its bandwidth digital conversion is of Figure 4. programming Considering switches. 300 Hz bandwidth of and the ECG signal [26], the LPF is adjusted done by the 12-bit AS-SAR ADC. In this way, a two sensor interface for the touch and the ECG are through programming switches. After the BRF and additional amplification, its digital conversion is supported by utilizing a single amplifier-based readout circuit, area are doneeffectively by the 12-bit AS-SAR ADC. In this way, a two sensor interface forthus, the minimizing touch and their the ECG and power consumption. The capacitive amplifier is designed to support rail-to-rail operation and effectively supported by utilizing a single amplifier-based readout circuit, thus, minimizing theirtoarea be fully differential for noise immunity. Its detailed schematics with common-mode feedback is and power consumption. The capacitive amplifier is designed to support rail-to-rail operation and to shown in Figure 4b, where schematics of the LPF and the BRF are also included.

be fully differential for noise immunity. Its detailed schematics with common-mode feedback is shown in Figure 4b, where schematics of the LPF and the BRF are also included. 3.3. Design of 12-Bit Alternate-Sampling SAR ADC Most sensorAlternate-Sampling interfaces are single-ended so that they are weak to various noises and fluctuations, 3.3. Design of 12-Bit SAR ADC

even though the following ADC circuit is fully differential. Figure 5 shows circuit-level processMost sensor interfaces single-ended so effective that theynumber are weak to various anddepending fluctuations, corner simulation results are about how much the of bits (ENOB)noises degrades

even though the following ADC circuit is fully differential. Figure 5 shows circuit-level process-corner

Sensors 2017, 17, 759

7 of 13

Sensors 2017, 17, 759 7 of 14 simulation results about how much the effective number of bits (ENOB) degrades depending on the input-stage structure of either single-ended or differential input when common-mode noises at SAR on the input-stage structure of either single-ended or differential input when common-mode noises ADCat inputs become stronger. conventional single-ended input structure is easily SAR ADC inputs becomeWhile stronger. While conventional single-ended input structure is corrupted, easily differential input structure is very immune to the common-mode noises. This is one of the motivations corrupted, input structure is very immune to the common-mode noises. This is one the Sensors 2017,differential 17, 759 7 of of 14 to propose the AS technique generates differential signals atsignals single-ended sensorsensor interfaces. motivations to propose the which AS technique which generates differential at single-ended onbenefit the input-stage structure of either or differential input when common-mode noisesa2.12-bit Another of the AS is theof capability error correction as shown in Figure 2. Therefore, interfaces. Another benefit the ASsingle-ended isofthe capability of error correction as shown in Figure at SAR ADC inputs become stronger. While conventional single-ended input structure is Therefore, 12-bit in AS-SAR ADC is used in high-resolution interfaces of touch, ECG,iteasily and BI, AS-SAR ADC isa used high-resolution sensor interfaces ofsensor touch, ECG, and BI, while is replaced corrupted, differential input structure is very immune to the common-mode noises. This is one of the while it is replaced with the XDC in cases of the other sensor interfaces. with the XDC in cases of the other sensor interfaces.

motivations to propose the AS technique which generates differential signals at single-ended sensor interfaces. Another benefit of the AS is the capability of error correction as shown in Figure 2. Therefore, a 12-bit AS-SAR ADC is used in high-resolution sensor interfaces of touch, ECG, and BI, while it is replaced with the XDC in cases of the other sensor interfaces.

Figure 5. Spectrum analysis of successive approximation register (SAR) analog-to-digital converters

Figure 5. Spectrum analysis of successive approximation register (SAR) analog-to-digital converters (ADC) architectures with random noise. (ADC) architectures with random noise.

Figure 6 shows a 12-bit fully differential AS-SAR ADC, where binary-weighted C-DACs adopt Figure 5. Spectrum analysis of successive approximation register (SAR) analog-to-digital converters

Figure 6capacitor shows astructure 12-bit fully differential where C-DACs adopt the split for smaller areas AS-SAR and betterADC, matching. Thebinary-weighted C-DAC array is split into two (ADC) architectures with random noise. identical sub-DACs by the attenuation capacitor is given by C-DAC array is split into two the split capacitor structure for smaller areas and (𝐶 better matching. The 𝐴𝑇𝑇 ) which Figure 6 shows a 12-bit fully differential AS-SAR ADC, binary-weighted C-DACs adopt identical sub-DACs by the attenuation capacitor (C ATT𝑜𝑓 ) which is given by 𝑇𝑜𝑡𝑎𝑙 𝑐𝑎𝑝𝑎𝑐𝑖𝑡𝑎𝑛𝑐𝑒 𝑀𝑆𝐵where 𝑝𝑎𝑟𝑡𝑠 = (2) 𝐴𝑇𝑇 smaller the split capacitor structure𝐶for and better𝑜𝑓matching. The C-DAC array is split into two 𝑇𝑜𝑡𝑎𝑙areas 𝑐𝑎𝑝𝑎𝑐𝑖𝑡𝑎𝑛𝑐𝑒 𝐿𝑆𝐵 𝑝𝑎𝑟𝑡𝑠 identical sub-DACs by the attenuation capacitor (𝐶𝐴𝑇𝑇 ) which is given by Total capacitance o f MSB parts

C ATT = 𝐶𝐴𝑇𝑇 =

Total capacitance o f LSB 𝑇𝑜𝑡𝑎𝑙 𝑐𝑎𝑝𝑎𝑐𝑖𝑡𝑎𝑛𝑐𝑒 𝑜𝑓 𝑀𝑆𝐵 𝑝𝑎𝑟𝑡𝑠parts 𝑇𝑜𝑡𝑎𝑙 𝑐𝑎𝑝𝑎𝑐𝑖𝑡𝑎𝑛𝑐𝑒 𝑜𝑓 𝐿𝑆𝐵 𝑝𝑎𝑟𝑡𝑠

(2)

(2)

Figure 6. Binary weighted split capacitive digital-to-analog converters (C-DACs).

In case of 12-bit implementation, 𝐶𝐴𝑇𝑇 becomes (64/63)* 𝐶𝑢𝑛𝑖𝑡 . In Figure 6, the right side of 𝐶𝐴𝑇𝑇 Figure 6. Binary weighted split capacitive digital-to-analog converters (C-DACs). 6. Binary weighted capacitive is of MSBFigure (most significant bit) part split and the left sidedigital-to-analog is of LSB (least converters significant (C-DACs). bit) part. If the 𝐶𝐴𝑇𝑇 value In is exactly the same as in (2), series combination of the LSB part and 𝐶 equivalent unit 𝐴𝑇𝑇 case of 12-bit implementation, 𝐶𝐴𝑇𝑇 becomes (64/63)* 𝐶𝑢𝑛𝑖𝑡 . In Figure 6, theisright side of to 𝐶𝐴𝑇𝑇 capacitance (𝐶 ). However, its real implementation is hard to match exactly because of parasitic 𝑢𝑛𝑖𝑡 In case of 12-bit implementation, C becomes (64/63)* C . In Figure 6, the right side of C ATT is of MSB (most significant bit) part and ATTthe left side is of LSB (least unit significant bit) part. If the 𝐶𝐴𝑇𝑇 effects. So, its revised equivalent capacitance with parasitic effects is given by value is exactly the same as in (2), series combination of the LSB part and 𝐶 is equivalent to unit is of MSB (most significant bit) part and the left side is of LSB (least significant bit) part. If the C 𝐴𝑇𝑇

ATT

capacitance (𝐶𝑢𝑛𝑖𝑡 ). However, real implementation is hard to match because of parasiticto unit value is exactly the same as in (2),itsseries combination of the LSB part exactly and C ATT is equivalent effects. So, its revised equivalent capacitance with parasitic effects is given by capacitance (Cunit ). However, its real implementation is hard to match exactly because of parasitic effects. So, its revised equivalent capacitance with parasitic effects is given by

Sensors 2017, 17, 759 Sensors 2017, 17, 759

8 of 13 8 of 14

(𝐶𝐴𝑇𝑇 + 𝐶𝑃1 ) × 2k𝑘 ∙ 𝐶𝑢𝑛𝑖𝑡 (3) 𝐶𝐸𝑞 = (𝑘C ATT + CP1 ) × 2 ·Cunit (3) CEq = k2 ∙ 𝐶𝑢𝑛𝑖𝑡 + 𝐶𝑃1 + 𝐶𝐴𝑇𝑇 + 𝐶𝑃𝐵 2 ·Cunit + CP1 + C ATT + CPB where C𝐶𝑃𝐵 isis parasitic parasitic capacitance capacitance between between the the MSB MSB part part and and the the LSB LSB part part and and C𝐶𝑃1 isis from from the the where PB P1 switched-capacitorarray arrayand anditsits layout routings in the 𝐶𝑃1 critically degrades the switched-capacitor layout routings in the LSBLSB part.part. SinceSince CP1 critically degrades the ADC ADC accuracy, its effect is minimized by optimizing the 𝐶 value through post-layout simulations. 𝐴𝑇𝑇 accuracy, its effect is minimized by optimizing the C value through post-layout simulations. ATT

4. Measurement Measurement Results Results & & Discussions Discussions 4. The proposed was fabricated in ain 0.18 metal The proposed multi-sensor multi-sensorROIC ROICprototype prototype was fabricated a μm 0.18 complementary µm complementary oxide oxide semiconductor (CMOS), and Figure shows microphotograph whose chip size is 2.9 size mm metal semiconductor (CMOS), and 7a Figure 7aits shows its microphotograph whose chip 2 , oscillator-based × 2.9 mm.× The 1.9 mm the ROIC, the amplifier-based ROIC, and is 2.9 mm 2.9core mm.area Theiscore area2,iswhere 1.9 mm where the oscillator-based ROIC, the amplifier-based the AS-SAR ADC occupy 0.28 mm2, 0.28 0.67 mm mm22,, 0.67 0.95 mm mm22,, 0.95 respectively. Figure 7b shows ROIC, and the AS-SAR ADC occupy mm2 , respectively. Figure a7bdisplayshows multi-sensor prototypeprototype and its measurement setup, where 8.26-inch 32 × 8 channel abased display-based multi-sensor and its measurement setup,anwhere an 8.26-inch 32 × 8 capacitive TSP was utilized. For implementation of fiveof heterogeneous sensing functions, the channel capacitive TSP was utilized. For implementation five heterogeneous sensing functions, reconfigurable multi-sensor interface was configured as the TSP-based detection mode and the the reconfigurable multi-sensor interface was configured as the TSP-based detection mode and the environmentaldetection detectionmode. mode. Figure 8 shows measurement results the TSP-based detection environmental Figure 8 shows measurement results of theofTSP-based detection mode mode for the touch, the ECG, and the BI. The touch function was verified to achieve a signal-to-noise for the touch, the ECG, and the BI. The touch function was verified to achieve a signal-to-noise ratio ratio (SNR) of 44.36 dB through finger-touch operation as shown in Figure 8a. The measured (SNR) of 44.36 dB through finger-touch operation as shown in Figure 8a. The ECGECG waswas measured by by placing hands onto two different half-planes the TSPs,and anditsitsmeasured measuredECG ECGwaveform waveformisis placing twotwo hands onto two different half-planes of of the TSPs, shown in in Figure Figure8b. 8b. For For the the BI BI measurement, measurement, all all TSP TSP pixels pixels were were utilized utilized to to maximize maximize the the capacitive capacitive shown amplifiergain, gain,and andFigure Figure8c 8cshows showsmeasured measuredfrequency frequencycharacteristics. characteristics. amplifier Figure 9a gives the ROIC characteristic responses in the amplifier-based amplifier-based configuration configuration when when Figure 9a gives the ROIC characteristic responses in the supply voltage and temperature are swept from 1.6 V to 2.0 V and from −12 ℃ to 85 ℃, respectively. supply voltage and temperature are swept from 1.6 V to 2.0 V and from −12 to 85 , respectively. Itsmeasured measuredSNR, SNR,especially especiallyin inthe the touch touch sensing sensing mode, mode, improved improvedslightly slightly with with respect respect to to increments increments Its of supply and temperature. This improvement is supposed to be caused by gain increments of the of supply and temperature. This improvement is supposed to be caused by gain increments of differential amplifier whose process-corner simulation results are given in Figure 10. Figure 9b shows the differential amplifier whose process-corner simulation results are given in Figure 10. Figure 9b the power-supply rejection ratio (PSRR) of the of amplifier-based readout circuit.circuit. Due toDue its to fullyshows the power-supply rejection ratio (PSRR) the amplifier-based readout its differential implementation, it achieved the PSRR of 71 dB at 60 Hz where a lot of power-supply noise fully-differential implementation, it achieved the PSRR of 71 dB at 60 Hz where a lot of power-supply appears. The XDC mode formode capacitive and resistive was functionally verified by noise appears. Thedetection XDC detection for capacitive andsensors resistive sensors was functionally utilizing external resistors and capacitors. Figure 11 shows two measured sensing characteristics by verified by utilizing external resistors and capacitors. Figure 11 shows two measured sensing utilizing the single oscillator-based XDC circuit, where it supports the capacitive sensing range from characteristics by utilizing the single oscillator-based XDC circuit, where it supports the capacitive 0 pF to 27 pF and the0 resistive of the fromresistive 270 kΩ range to 1 MΩ. The effect of supply-voltage sensing range from pF to 27 range pF and of from 270 kΩ to 1 MΩ. The variation effect of on the capacitive and resistive sensor characteristics are also included together, which means that an supply-voltage variation on the capacitive and resistive sensor characteristics are also included together, additional supply-voltage detection circuit might calibrate the corresponding characteristic which means that an additional supply-voltage detection circuit might calibratesensor the corresponding variation. sensor characteristic variation.

(a)

(b)

Figure7. 7. Proposed Proposed reconfigurable reconfigurable readout readout integrated integrated circuit circuit (ROIC) (ROIC) prototype: prototype: (a) (a) microphotograph; microphotograph, Figure (b)measurement measurementsetup. setup. (b)

Sensors 2017, 17, 759 Sensors 2017, 17, 759

9 of 13 9 of 14

Sensors 2017, 17, 759

9 of 14

(a)

(b)

(c)

(a) (b) (c) Figure 8. panel (TSP)-based sensing functions: (a) touch, (b) Figure 8. Measurement Measurementresults resultsofoftouch touchscreen screen panel (TSP)-based sensing functions: (a) touch; electrocardiogram (ECG), andand (c) body impedance. (b) electrocardiogram (ECG); body impedance. Figure 8. Measurement results of(c) touch screen panel (TSP)-based sensing functions: (a) touch, (b) electrocardiogram (ECG), and (c) body impedance.

(a)

(b)

(b) Figure 9. (a) Measured(a) ROIC signal-to-noise ratios (SNRs) over temperature and supply-voltage variations, (b) ROIC power-supply rejection ratio (PSRR) (simulated). Figure 9. 9. (a) signal-to-noise ratios (SNRs) over over temperature temperature and and supply-voltage supply-voltage Figure (a) Measured Measured ROIC ROIC signal-to-noise ratios (SNRs) variations, (b) (b) ROIC power-supply power-supply rejection rejection ratio ratio (PSRR) (simulated). (simulated). variations; The AS-SAR ROIC ADC, which is commonly used (PSRR) for the TSP-based sensors, was integrated together in the ROIC, but post-processing algorithms of the interpolation and error correction were The AS-SAR ADC, which is commonly used for the TSP-based sensors, was integrated together The AS-SAR which is commonly for the the TSP-based sensors, integrated together in implemented in ADC, the MATLAB. Figure 12used compares measured fast was Fourier transform (FFT) in the ROIC, but post-processing algorithms of the interpolation and error correction were the ROIC, but post-processing algorithms of the interpolation and error correction were implemented spectrums before and after applying the proposed alternate-sampling and error-correction scheme. implemented in the MATLAB. Figure 12 compares the measured fast Fourier transform (FFT) in MATLAB. Figure 12 compares the measured fast Fourier transformfrom (FFT) spectrums before Thethe measured signal-to-noise and distortion ratio (SNDR) was improved 45.2 dB to 71.17 dB, spectrums before and after applying the proposed alternate-sampling and error-correction scheme. and the aftermeasured applyingeffective the proposed alternate-sampling andfrom error-correction scheme. number of bits (ENOB) was 7.25 bits to 11.53 bits. The The measured The measured signal-to-noise and distortion ratio (SNDR) was improved from 45.2 dB to 71.17 dB, signal-to-noise and distortion (SNDR) was improved improved from 45.2 toto 71.17 dB,dB. and spurious-free dynamic range ratio (SFDR) was also 49.99dBdB 83.03 Inthe thismeasured AS-SAR and the measured effective number of bits (ENOB) was from 7.25 bits to 11.53 bits. The measured effective of utilized bits (ENOB) from 7.25 bits to 11.53 bits.devices The measured dynamic ADC, thenumber C-DACs 70 fFwas metal-insulator-metal (MIM) as unit spurious-free capacitors to meet the spurious-free dynamic range (SFDR) was also improved from 49.99 dB to 83.03 dB. In this AS-SAR range (SFDR) was also improved from 49.99 dB to 83.03 dB. In this AS-SAR ADC, the C-DACs utilized kT/C noise requirement. ADC, the C-DACs utilized 70 fF metal-insulator-metal (MIM) devices as unit capacitors to meet the 70 fF metal-insulator-metal (MIM) devices as unit capacitors to meet the kT/C noise requirement. kT/C noise requirement.

Sensors 2017, 17, 759 Sensors 2017, 17, 759

10 of 13

10 of 14

Sensors 2017, 17, 759 Sensors 2017, 17, 759

10 of 14 10 of 14

(a) (b) (b) (a) (b) Figure 10. Fully-differential amplifier characteristic over process-voltage-temperature (PVT) Figure 10. 10. Fully-differential amplifier characteristic over process-voltage-temperature (PVT) variations Figure Fully-differential amplifier characteristic over process-voltage-temperature (PVT) Figure 10. Fully-differential amplifier characteristic over process-voltage-temperature (PVT) variations (simulated): (a) gain (b) phase margin. (simulated): (a) gain (b) phase margin. variations (simulated): (a) gain (b) phase margin. variations (simulated): (a) gain (b) phase margin.

(a)

(a) (a) (a)

(b) (b) (b)

Figure 11. Measured characteristics of X-to-digital converters (XDC)-based readout circuit: Figure 11. and Measured characteristics of X-to-digital converters (XDC)-based readout circuit: (a) capacitive (b) resistive. Figure Measured characteristics of Figure 11.11. Measured characteristics of X-to-digital X-to-digital converters converters (XDC)-based (XDC)-basedreadout readoutcircuit: circuit: (a) capacitive and (b) resistive.

capacitive and resistive. (a) (a) capacitive and (b)(b) resistive. Figure 13 shows the power breakdown of sub-blocks in the proposed reconfigurable ROIC, 13 shows the power breakdown of sub-blocks proposed reconfigurable ROIC, whereFigure their average power consumption is given depending in onthe their operational configuration. In Figure 13 shows the power breakdown of sub-blocks in the proposed reconfigurable ROIC, where average power consumption isthe given depending on their operational configuration. In Figure 13 shows the power breakdown ofoscillator sub-blocks proposed reconfigurable ROIC, the casetheir of the oscillator-based operations, and in thethe single-ended amplifier consume where theirofaverage power consumption is given depending on their operational configuration. the case the26.23 oscillator-based operations, the depending oscillator and single-ended amplifier consumeInInthe where their average power is given onthe their operational configuration. 126.18 μW and μW,consumption respectively. The digital conversion requires about 73 μW. In the amplifierthe126.18 case of operations, the oscillator and the single-ended amplifier consume μWthe andoscillator-based 26.23 respectively. digital conversion requires 73 μW. In the the amplifiermode,μW, differential amplifiers and filters for better noiseabout immunity increase ROIC case based of theoperation oscillator-based operations, theThe oscillator and the single-ended amplifier consume 126.18 µW 126.18 μW and 26.23 μW, respectively. The digital conversion requires about 73 μW. In the amplifierbased operation mode, differential amplifiers and filters for better noise immunity increase the consumption. Total power consumption is 0.23requires mW for the oscillator-based mode andROIC 1.09 and power 26.23 µW, respectively. The digital conversion about 73 µW. In the amplifier-based based operation mode, differential amplifiers andisfilters for better immunityfor increase the1.09 ROIC power consumption. Totalmode, power consumption 0.23 mW for thenoise oscillator-based mode and mW for the amplifier-based resulting in optimization of the power consumption its operational operation mode, differential amplifiers and filters for better noise immunity increase the ROIC power consumption. Total power consumption is 0.23 mW forpower the oscillator-based mode and power 1.09 mW for the amplifier-based mode, resulting in optimization of the consumption for its operational modes. consumption. Total power consumption is 0.23 mW for the oscillator-based mode and 1.09 mW for the mW for the amplifier-based mode, resulting in optimization of the power consumption for its operational modes.

amplifier-based mode, resulting in optimization of the power consumption for its operational modes. modes.

(a) (a)

(a) Figure 12. Cont.

Sensors 2017, 17, 759

11 of 13

Sensors 2017, 17, 759

11 of 14

Sensors 2017, 17, 759

11 of 14

(b) (b) Figure 12. Measured 262144-point fast Fourier transform (FFT) spectrums at 1.777 kHz input (a) with

Figure 12.12.Measured fastFourier Fourier transform (FFT) spectrums at 1.777 kHz(a) input Figure Measured262144-point 262144-point fast transform (FFT) spectrums at 1.777 kHz input with(a) with (b) without alternate sampling and error correction scheme. without alternatesampling sampling and scheme. (b)(b) without alternate anderror errorcorrection correction scheme.

(a)(a)

(b) (b)

Figure 13. Power breakdown of the sub-blocks: (a) oscillator-based circuits and (b) amplifier-based circuits.

Figure Power breakdownofofthe thesub-blocks: sub-blocks: (a) (b)(b) amplifier-based circuits. Figure 13.13. Power breakdown (a) oscillator-based oscillator-basedcircuits circuitsand and amplifier-based circuits.

This works, which areare summarized in Table 1. The proposed This work workwas wascompared comparedwith withrecent recent works, which summarized in Table 1. The proposed This workROIC was with recent works, which are summarized insensing Table 1.functions The proposed multi-sensor was verified to to support fivefive heterogeneous multi-sensor ROICcompared wasexperimentally experimentally verified support heterogeneous sensing functions multi-sensor was experimentally verified totwo support five heterogeneous sensing functions while previous works provide one or or two sensing functions. Thanks to the proposed while most mostROIC previous works provideonly only one sensing functions. Thanks to the proposed reconfigurable ROIC structure, wide ranges of sensor capacitance and resistance are simultaneously while most previous works provide only one or two sensing functions. Thanks to the proposed reconfigurable ROIC structure, wide ranges of sensor capacitance and resistance are simultaneously supported, the itself provides competitive performance of the SNRSNR withsimultaneously smaller reconfigurable ROIC structure, wide ranges of sensor capacitance and resistance are supported, and and theTSP TSPinterface interface itself provides competitive performance of the with smaller area and power consumption. In the proposed reconfigurable ROIC structure, three TSP-based supported, the consumption. TSP interface In itself performance of thethree SNR TSP-based with smaller area and and power the provides proposedcompetitive reconfigurable ROIC structure, sensing circuits were implemented to reduce the area and the power consumption by selectively area and power In the ROICconsumption structure, three TSP-based sensing circuitsconsumption. were implemented to proposed reduce the reconfigurable area and the power by selectively reusing the single readout circuit for all 32 × 8 TSP pixels and also by utilizing the TSPs as capacitive reusing the single readout circuit for all 32 × 8 TSP pixels and also by utilizing the TSPs as capacitive sensing circuits were implemented to reduce the area and the power consumption by selectively electrodes. The ADC was designed to have 12-bit resolution which is sufficient for most sensing electrodes. The readout ADC was designed to32 have resolution which is sufficient for most reusing the single circuit × 812-bit TSP pixels and also by utilizing theimplemented TSPs as sensing capacitive functions, including the ECG and for the all TSPs. The environmental sensing functions were functions,The including the ECG and thetoTSPs. The environmental sensing functions were implemented electrodes. ADC was designed have 12-bit resolution which is sufficient for most by utilizing the same readout circuit without additional area, and its low-power characteristic wassensing by utilizing the same readout circuit without area, and its low-power was functions, theXDC ECG and thewhich TSPs. Theadditional environmental sensing functionscharacteristic were implemented achievedincluding through the operation does not need to activate the ADC. achieved through the XDC operation which does not need to activate the ADC. by utilizing the same readout circuit without additional area, and its low-power characteristic was Table 1. Performance summary and comparison with recent works.

achieved through the Table XDC 1. operation which doesand notcomparison need to activate the works. ADC. Performance summary with recent Parameter Parameter Process Process Supported sensor types sensor Supported Parameter types

This Work [1] ThisµWork 0.18 m 0.18 µ[1] m Table 1. Performance summary and 0.18 µ mbody 0.18 µ m TSP, R/C type, TSP impedance TSP, R/C (BI), type,ECG body TSP This Work [1] impedance (BI), ECG 1.163 (TSP/ECG + Process 0.18 µm 0.18 µm Power consumption 1.163ADC) (TSP/ECG Supported sensor TSP, R/C type, body + 6.26 (TSP) TSP (mW) 0.202(BI), (XDC) Powertypes consumption impedance ADC) ECG 6.26 (TSP) 0.271 (BI +(XDC) ADC) (mW) 0.202

Power consumption (mW)

Area (mm2) 2 ) 2) Area (mm Area (mm

1.163 (TSP/ECG + ADC) 0.202 0.271(XDC) (BI + ADC) 0.67 0.271 (BI(TSP/ECG) + ADC)

6.26 (TSP)

0.057 (XDC)

(TSP/ECG) 0.670.67 (TSP/ECG) 0.223 (BI) 0.057 (XDC) 0.057 (XDC) 0.95 (BI) (ADC) 0.2230.223 (BI) 0.950.95 (ADC) (ADC)

2.2 2.2

2.2

[9] [9] 0.18 µ m comparison with 0.18 µ m BI, ECG BI, ECG [9]

[10] [11] [10] 0.16 µ m 0.18 µ m[11] recent works. 0.16 µ m 0.18 µ m C type R type C type R type [10] [11] 0.014 0.0017 0.18 µm 0.16 µm 0.014 0.18 µm (capacitance(resistance0.0017 0.233 (3 channel BI to-digital to-digital (capacitance(resistanceBI, ECG C type R type + ECG ) 0.233 (3 channel BI converter to-digital converter to-digital + ECG ) 0.014 0.0017 (CDC)) (RDC)) converter converter 0.233 (3 channel (capacitance-to-digital (resistance-to-digital (estimated per BI8.17 + ECG ) (CDC)) (RDC)) converter (CDC)) converter (RDC)) channel) 8.17 (estimated per 8.17 (estimated 49 (ROIC + micro 0.05 1.23 per channel)channel) controller unit 49 (ROIC 49 + micro 1.23 (ROIC + micro 0.05 0.05 1.23 controller(MCU)) unit controller unit (MCU)) (MCU))

Application

Bio-signal acquisition, TSP, R & C type sensor

TSP

Bio-signal acquisition

C type sensor

R type sensor

Supply (V)

1.8

2.1–3.3

1.2

1

1.2/0.6

Sensors 2017, 17, 759

12 of 13

Table 1. Cont. Parameter

This Work

[1]

[9]

[10]

[11]

Resolution (bits)

12

N.A.

13.5

13.1 (CDC)

10

R, C sensing Range (Ω), (F)

R: 270 k–1 M C: 0 p–27 p

N.A.

N.A.

C: 0–8 p

R: 10 k–10 M

TSP channel

TX: 32 RX: 8

TX: 12 RX: 8

N.A.

N.A.

N.A.

Frame rate (Hz)

240

50–6400

N.A.

N.A.

N.A.

TSP SNR (dB)

44.36

40 @ oversampling ratio (OSR) 32

N.A.

N.A.

N.A.

Body impedance frequency range (Hz)

50 k–750 k

N.A.

1–20 M

N.A.

N.A.

5. Conclusions This paper presented a reconfigurable readout integrated circuit with the following two operation modes: amplifier-based and the oscillator-based. Its system prototype of the display-based multi-sensor interface was implemented and experimentally verified to support five sensor types: touch, ECG, BI, resistive, and capacitive environmental sensors. By making sophisticated changes to the TSP interface configuration, the three sensory interfaces of touch, ECG, and BI were supported. The oscillator-based operation mode was utilized to support resistive and capacitive environmental sensors and stretchable interfaces. The AS scheme, embedding differential conversion, signal doubling, and error correction, was proposed to achieve better effective resolution and noise immunity against inherent human-touch operations. A 12-bit AS-SAR ADC was integrated in the ROIC and the proposed AS scheme was experimentally verified to provide about 4-bit improvement of the ENOB. Other characteristics of SNDR and SFDR were improved from 45.42 dB to 71.17 dB and from 49.99 dB to 83.03 dB, respectively. This reconfigurable work was designed to provide heterogeneous multi-sensor interfaces in multiple applications of display systems, flexible electronics, and wearable devices. Potentially, the bio-signal interface can be utilized in healthcare devices, and also the XDC interface can be applied to various environmental monitoring systems. The AS scheme can correct two consecutive errors, and further correction capability will be provided through upcoming work. Acknowledgments: This work was supported by the Technology Innovation Program (10064058: Development of patch-type 7 modal multi-sensor devices and platform technology for wearable healthcare applications) funded by the Ministry of Trade, Industry and Energy, South Korea and also supported by the Basic Science Research Program through the National Research Foundation of Korea funded by the Ministry of Education under Grant 2016R1D1A1B03930404. Author Contributions: K.P. and J.J.K. conceived and designed the multi-sensor interface architecture; K.P., S.M.K., and W.-J.E. implemented the ROIC and performed its system experiments; K.P. and J.J.K. wrote the paper. Conflicts of Interest: The authors declare no conflict of interest.

References 1. 2. 3. 4. 5. 6.

Park, J.-E.; Lim, D.-H.; Jeong, D.-K. A reconfigurable 40-to-67 dB SNR, 50-to-6400 Hz frame-rate, column-parallel readout IC for capacitive touch-screen panels. IEEE J. Solid State Circuit 2014, 49, 2305–2318. [CrossRef] Kim, H.-S.; Han, K.-Y. High-SNR capacitive multi-touch sensing technique for AMOLED display panels. IEEE Sens. J. 2016, 16, 859–860. [CrossRef] Kwon, K.; Kim, S.-W.; Bien, F.; Kim, J.J. A fully-differential correlated double sampling readout circuit for mutual-capacitance touch screens. J. Semicond. Technol. Sci. 2015, 15, 349–355. [CrossRef] Yusefpor, M.; Bussat, J.M.; Lyon, B.B. Fingerprint Sensor in an Electronic Device. U.S. Patent Application No 14/451,076; U.S. Patent Publication No 2015/0036065 A1, 5 February 2015. Eom, W.-J.; Kim, S.-W.; Park, K.; Bien, F.; Kim, J.J. Multi-purpose fingerprint readout circuit embedding physiological signal detection. J. Semicond. Technol. Sci. 2016, 16, 793–799. [CrossRef] Martinez-Millana, A.; Bayo-Monton, J.L.; Lizondo, A.; Fernandez-Llatas, C. Evaluation of Google Glass technical limitations on their integration in medical systems. Sensors 2016, 16, 2142. [CrossRef] [PubMed]

Sensors 2017, 17, 759

7. 8. 9.

10.

11.

12. 13. 14.

15.

16. 17. 18. 19. 20.

21. 22. 23. 24. 25.

26.

13 of 13

Patel, S.; Park, H.; Bonato, P.; Chan, L.; Rodgers, M. A review of wearable sensors and systems with application in rehabilitation. J. Neuro Eng. Rehabil. 2012. [CrossRef] [PubMed] Meijer, G.; Pertijs, M.; Makinwa, K. Smart Sensor Systems: Emerging Technologies and Applications, 1st ed.; Wiley: West Sussex, UK, 2014; pp. 1–14. Helleputte, N.V.; Konijnenburg, M.; Pettine, J.; Jee, D.-W.; Kim, H.; Morgado, A.; Wegberg, R.V.; Torfs, T.; Mohan, R.; Breeschoten, A.; et al. A 345 µW multi-sensor biomedical SoC with bio-impedance, 3-channel ECG, motion artifact reduction, and integrated DSP. IEEE J. Solid State Circuit 2015, 50, 230–244. [CrossRef] He, Y.; Chang, Z.-Y.; Pakula, L.; Shalmany, S.H.; Pertijs, M. A 0.05 mm2 1V Capacitance-to-Digital Converter Based on Period Modulation. In Proceedings of the IEEE International Solid-State Circuits Conference, Digest of Technical Papers, San Fransisco, CA, USA, 22–26 February 2015; pp. 486–487. Choi, M.; Gu, J.; Blaauw, D.; Sylvester, D. Wide Input Range 1.7 µW 1.2 kS/s Resistive Sensor Interface Circuit with 1 cycle/sample Logarithmic Sub-Ranging. In Proceedings of the Symposium on VLSI Circuits, Digest of Technical Papers, Kyoto, Japan, 15–19 June 2015; pp. 330–331. Kyle, U.G.; Bosaeus, I.; Lorenzo, A.D.D.; Deurenberg, P. Bioelectrical impedance analysis—Part I: Review of principles and methods. Clin. Nutr. 2004, 23, 1226–1243. [CrossRef] [PubMed] Abad, D.G. Development of a Capacitive Bioimpedance Measurement System. Ph.D. Thesis, RWTH Aachen University, Aachen, Germany, 2009. Bouchaala, D.; Shi, Q.; Chen, X.; Kanoun, O.; Derbel, N. Comparative study of voltage controlled current sources for biompedance measurements. In Proceedings of the International Multi-Conference on Systems, Signals and Devices, Hammamet, Tunisia, 18–21 March 2013. Ciuti, G.; Ricotti, L.; Menciassi, A.; Dario, P. MEMS sensor technologies for human centred applications in healthcare, physical activities, safety and environmental sensing: A review on research activities in Italy. Sensors 2015, 15, 6441–6468. [CrossRef] [PubMed] Xu, R.; Liu, B.; Yuan, J. Digitally calibrated 768-kS/s 10-b minimum-size SAR ADC array with dithering. IEEE J. Solid State Circuit 2012, 47, 2129–2140. [CrossRef] Lin, Z.; Wu, J.; Chen, C. Digital-domain dual-calibration for single-ended successive approximation register ADCs. Electron. Lett. 2015, 51, 1161–1163. [CrossRef] Chen, Y.-J.; Chang, K.-H.; Hsieh, C.-C. A 2.02–5.16 fJ/conversion step 10 bit hybrid coarse-fine SAR ADC with time-domain quantizer in 90 nm CMOS. IEEE J. Solid State Circuit 2016, 51, 357–364. Harpe, P.; Cantatore, E.; Roermund, A.V. A 10 b/12 b 40 kS/s SAR ADC with data-driven noise reduction achieving up to 10.1b ENOB at 2.2 fJ/conversion-step. IEEE J. Solid State Circuit 2013, 48, 3011–3018. [CrossRef] Chan, C.-H.; Zhu, Y.; Ho, I.-M. A 0.011 mm2 60 dB SNDR 100MS/s reference error calibrated SAR ADC with 3pF decoupling capacitance for reference voltages. In Proceedings of the IEEE Asian Solid-State Circuits Conference, Toyama, Japan, 7–9 November 2016; pp. 145–148. Tsai, J.-H.; Wang, H.-H.; Yen, Y.-C. A 0.003 mm2 10 b 240 MS/s 0.7 mW SAR ADC in 28 nm CMOS with digital error correction and correlated-reversed switching. IEEE J. Solid State Circuit 2015, 50, 1382–1398. [CrossRef] Enz, C.C.; Temes, G.C. Circuit techniques for reducing the effects of op-amp imperfections: Autozeroing, correlated double sampling, and chopper stabilization. Proc. IEEE 1996, 84, 1584–1614. [CrossRef] Chang, D.-J.; Kim, W.; Seo, M.-J. Normalized-full-scale-referencing digitial-domain linearity calibration for SAR ADC. IEEE Trans. Circuits Syst. I 2017, 64, 322–332. [CrossRef] An, Y.-J.; Jung, D.-H.; Ryu, K.; Yim, H.S.; Jung, S.-O. All-digital on-chip process sensor using ratioed inverter-based ring oscillator. IEEE Trans. VLSI Syst. 2016, 24, 3232–3242. [CrossRef] Shin, H.; Ko, S.; Jang, H.; Yun, I.; Lee, K. A 55 dB SNR with 240 Hz frame scan rate mutual capacitor 30 × 24 touch-screen panel read-out IC using code-division multiple sensing technique. In Proceedings of the IEEE International Solid-State Circuits Conference, Digest of Technical Papers, San Fransisco, CA, USA, 17–21 February 2013; pp. 388–389. Song, H.; Park, Y.; Kim, H.; Ko, H. Fully integrated biopotential acquisition analog front-end IC. Sensors 2015, 15, 25139–25156. [CrossRef] [PubMed] © 2017 by the authors. Licensee MDPI, Basel, Switzerland. This article is an open access article distributed under the terms and conditions of the Creative Commons Attribution (CC BY) license (http://creativecommons.org/licenses/by/4.0/).