31.3 nW, 0.5 V Bulk-Driven OTA for Biosignal Processing

Received 3 May 2023, accepted 21 May 2023, date of publication 24 May 2023, date of current version 12 June 2023. Digital Object Identifier 10.1109/ACCESS.2023.3279727 31.3 nW, 0.5 V Bulk-Driven OTA for Biosignal Processing MONTREE KUMNGERN 1, TOMASZ KULEJ2, AND FABIAN KHATEB 3,4,5 1Department of Telecommunications Engineering, School of Engineering, King Mongkut’s Institute of Technology Ladkrabang, Bangkok 10520, Thailand 2Department of Electrical Engineering, Częstochowa University of Technology, 42-201 Czestochowa, Poland 3Department of Microelectronics, Brno University of Technology, 601 90 Brno, Czech Republic 4Faculty of Biomedical Engineering, Czech Technical University in Prague, 272 01 Kladno, Czech Republic 5Department of Electrical Engineering, University of Defence, 662 10 Brno, Czech Republic Corresponding author: Montree Kumngern ([email protected]) This work was supported by the University of Defence Brno within the Organization Development Project VAROPS. ABSTRACT This paper presents a new extremely low-voltage low-power bulk-driven (BD) operational transconductance amplifier (OTA) realized for low frequency biosignal processing. The CMOS structure of the OTA utilizes bulk-driven and self-cascode techniques in the subthreshold region, supporting the operation with the supply voltage (VDD) as the threshold voltage (VTH) of a single MOS transistor, i.e., VDD =VTH =0.5 V, while offering nano power consumption (31.3 nW for 15 nA nominal setting current). Using the extremely low-voltage and low-power OTA in biosignal processing enables extending the lifetime of applications that are powered by battery or energy harvesting sources. The OTA has a 54.7 dB low frequency gain, 6.18 kHz gain bandwidth and 75◦phase margin at 15 pF load capacitance. The proposed OTA has been used to realize a bandpass filter (BPF) with adjustable gain for electrocardiogram (ECG) signal processing. The higher cutoff frequency of the BPF is adjustable electronically by a setting current and the BPF’s gain can be adjusted by capacitors value. The total harmonic distortion (THD) of the BPF is −53.56 dB, the input integrated input-referred voltage noise is 17.9 µVrms, the common mode rejection ratio (CMRR) is 75 dB and the power supply rejection ratio (PSRR) is 87.7 dB. The BPF was designed in the Cadence program using 0.18 µm CMOS technology from TSMC. The simulation results agree with the presented theory. INDEX TERMS Operational transconductance amplifier (OTA), bulk-driven, band-pass filter, low-voltage, low-power CMOS. I. INTRODUCTION Nowadays, systems such as portable, battery-operated, and self-powered electronics require extremely low-voltage operation capability and low-power consumption. In biomedical systems, such as data acquisition systems for electrocardiographic (ECG) signal processing, low-voltage and lowpower analog circuits are required at the front-end of the systems. Analog circuits, such as amplifiers and filters, are usually used to realize the front-end. Some examples include the operational amplifier (op-amp), transconductance amplifier (OTA), and the differential difference amplifier (DDA), which is based on complementary The associate editor coordinating the review of this manuscript and approving it for publication was Harikrishnan Ramiah . metal-oxide-semiconductor (CMOS) technologies. In the analog portion, reducing the voltage supply degrades the analog circuit performances, such as the dynamic range (DR), bandwidth and linearity. To maintain the required DR when the voltage supply is reduced, the input voltage swing should be increased. It is well-known that circuits capable of rail-torail input voltage swing are the best solution for maintaining the required DR when the circuits are operated with lowvoltage supply. The original technique for implementing the rail-to-rail input voltage swing of circuits is the use of differential pairs that are composed of both PMOS and NMOS transistors [1]. However, this technique is complex due to the additional differential pair, current branches and the circuitry used to maintain constant transconductance over the whole input voltage range. Moreover, this technique is not the most 56516 This work is licensed under a Creative Commons Attribution-NonCommercial-NoDerivatives 4.0 License. For more information, see https://creativecommons.org/licenses/by-nc-nd/4.0/ VOLUME 11, 2023 M. Kumngern et al.: 31.3 nW, 0.5 V Bulk-Driven OTA for Biosignal Processing suitable for circuits that operate with a low-power supply and low-power consumption. Therefore, non-conventional techniques, such as bulk-driven (BD) [2], floating-gate (FG) [3], quasi-floating gate (QFG) [4], or dynamic threshold voltage (DT) [5], which are suitable for circuits operating with lowsupply voltages, have been reported. Using these techniques, the input voltage swing of circuits can be extended due to the threshold voltage that can be reduced or removed from the signal path. Some low-voltage and low-power analog devices using non-conventional techniques are reported in literature, including the op-amp [6], [7], [8], [9], OTA [10], [11], [12], [13], [14], DDA [15], [16], [17], [18], [20], [39], [40], and second-generation current conveyor (CCII) [19], [20], [21], [22]. Low-voltage and low-power analog circuits can be applied to biosignal processing; analog filters are an example of this application [23], [24], [25], [26], [27], [28], [29], [30], [31], [32], [33], [34], [35], [36], [37], [38]. In [28], [29], [30], [31], [32], [33], [34], [35], [36], [37], and [38], BPFs suitable for applications in biomedical systems have been proposed. However, some of these BPFs suffer from highpower consumption [28], [29], [30], [31], [32], [35], [36], [37], [38], cannot control voltage gain [30], [31], [32], [33], [34], [35], [37], [38], or cannot control the higher and lower cutoff frequencies [29], [30], [31], [32], [33], [34], [35], [36], [37], [38]. In this work, the bulk-driven technique has been used to realize a new extremely low-voltage low-power OTA (BD-OTA) for low frequency biosignal processing. The OTA can work with a supply voltage of 0.5 V and a power of 31.3 nW for 15 nA nominal setting current. The proposed OTA has been used to realize a bandpass filter (BPF) with adjustable gain for ECG signal processing. The voltage gain and higher cutoff frequency can be controlled. II. PROPOSED CIRCUIT A. PROPOSED 0.5 V OTA The CMOS schematic of the 0.5-V OTA used to realize the BPF proposed in this work is shown in Fig. 1 (a). The structure was first proposed in [41] and verified experimentally in [42]. In both cases, it served as the input stage of a twostage operational amplifier. Here, the structure is used as a single-stage amplifier, with all transistors replaced by selfcomposite devices, as shown in Figs 1 (b) and 1 (c). The OTA shown in Fig. 1 (a) is in fact a single-stage nontailed differential amplifier. For simplicity, let us first consider the version with transistors M7and M8removed [43]. In such a case, transistors M1A with M2A and M1B with M2B form two current mirrors. The mirrors are biased with constant current sources based on transistors M3and M4. Note that the input differential signals V+and V−are applied to the bulk terminals of the input transistors. With V+=V−, the threshold voltages of M1A and M2A (M1B and M2B) are equal for every level of the input voltage. Therefore, neglecting the impact of drain-source conductances gds, the current gains of the above-mentioned current mirrors do not depend on the input common-mode signals. Note that in such a case, the drain currents of M1A and M1B are not only equal, but also remain constant and equal to the biasing currents ID3,4. This is a significant difference in comparison with pseudodifferential amplifiers, where currents of a differential stage highly depend on the common-mode level. The drain currents of M1B and M1A (ID1B and ID1A) are subtracted thanks to the unity-gain current mirror M5-M6; thus, the output current is equal to zero. Note that, neglecting second order effects, variations of VDD do not affect the drain currents of M1A and M1B. Thus, neglecting the impact of second order effects, the output signal of the OTA is insensitive to both the input common-mode level as well as variations of VDD. This results in relatively large values of the common-mode and power supply rejection ratios (CMRR and PSRR, respectively). When V+and V−are not equal, then the threshold voltages of M1and M2are different, thus producing a current difference between M1and M2and consequently a non-zero output current of the whole stage. Note that the value of this current is not limited by a tail current source and can approach significant values, much larger than the biasing currents ID3,4. This improves the slew-rate (SR) of the OTA. By removing the tail current source, which is used for biasing the differential amplifiers in conventional solutions, the minimum VDD can be decreased while still offering good CMRR and PSRR performances. This also increases the output swing of the OTA, which is nearly rail-to-rail. In addition, the input common-mode range of the amplifier is also rail-torail, due to the use of bulk-driven input transistors. By applying a single-stage OTA, a power-effective solution can be achieved as it eliminates the need for a second gain stage that usually consumes more power. However, sufficient voltage gain is problematic. In order to increase the voltage gain, two additional techniques have been applied. First, the voltage gain is increased by applying partial positive feedback (PPF), introduced by the cross-coupled transistors M7and M8. The transistors generate negative conductances at the drain/gate nodes of M2A and M2B, thus increasing the resulting resistance at these nodes, and consequently increasing the voltage gain from the input terminals to these nodes. Since transistors M1A and M1B are controlled with larger signals at their gates, the overall transconductance and the voltage gain of the OTA is increased. Referring to Fig. 1 (a), the DC voltage gain of the OTA can be expressed as [41]: Avo =gm gds1+gds6(1) where gmis the transconductance of the OTA, which in turn can be expressed as: gm=2gmb11+p/2 1−m+p(2) where m =gm7,8/gm2, p =(gds2 +gds7,8+gds3,4)/gm2, and gmb1 is the bulk transconductance of M1. Note that in a weak inversion region, the ratio of transconductances is equal to VOLUME 11, 2023 56517 M. Kumngern et al.: 31.3 nW, 0.5 V Bulk-Driven OTA for Biosignal Processing FIGURE 1. (a) 0.5 V bulk-driven OTA, (b) self-cascode PMOS, (c) self-cascode NMOS. the ratio of biasing currents, which is again equal to the W/L ratios of the corresponding MOS transistors. As we can conclude from (1) and (2), the voltage gain of the OTA depends on the coefficient m, which is equal to the ratio of transconductances gm7,8/gm2 that expresses the amount of positive feedback. The closer the coefficient m is to unity, the higher the voltage gain. The maximum value of this coefficient is limited by gm, the sensitivity to transistor mismatch. For instance, the gmsensitivity to parameter m can be expressed as [20]: Sgm m=m 1−m+p−m 1+m+gm1gm2≈m 1−m+p(3) Thus, the gmsensitivity increases as m tends to unity, i.e., with increasing voltage gain. The maximum value of m is also limited by the frequency properties of the OTA, since increasing the value of m increases the resistance seen at the gate/drain nodes of M2. This decreases the frequency of the parasitic pole associated with this node and can be expressed as [20]: ωp=1−m+p 1+m+gm1gm2·gm2 Cgs2(4) Thus, the frequency of the pole is approximately inversely proportional to the voltage gain Avo. The partial positive feedback also improves the CMRR and PSRR parameters of the OTA, which can be expressed as [41]: CMRR =21+p/2 p·(1 +m+p) (1 −m+p)(5) PSRR ≈21+p/2 p(1−m+p)(1 +gm2gmb2)(6) This is because PPF increases the differential voltage gain, without affecting the common-mode gain and the parasitic gain from VDD. The input-referred noise of the OTA can be described by the following equation [41]: ¯ v2 ni =N"8kT 3g2 mb1gm1+gm5,6 +1 g2 mb1fCOX Kpg2 m1 W1L1+Kng2 m5,6 W5,6L5,6!# (7) where: N=1 4 gm1gm2 (1+p/2)2"1+m+(1−m+p)2 gm1gm2#(8) where kis the Boltzmann constant, COX is the gate oxide capacitance per unit area, Kpand Knare the flicker noise constants for the pand n-channel transistors respectively, and the other symbols have their usual meaning or were defined earlier. It can be concluded from (7) and (8) that the input referred noise of the OTA is approximately equal to the noise of a conventional bulk-driven differential pair, biased with the same total current and the same sum of transistor channel areas [41], [42]. In order to further increase the DC voltage gain of the OTA in the proposed solution, each transistor (except for the biasing M3, M4and MB) was replaced with a so-called selfcascode composite transistor, as shown in Fig.2 (b) and (c). Such a solution increases the voltage gain without sacrificing the voltage swing at the output of the OTA. In a weak inversion region, with the |VDS|of the ‘‘bottom’’ triode operated transistor MStending to zero, the small signal transconductance of the self-composite transistor is approximately equal to the transconductance of the upper transistor MD. Thus, it is equal to the transconductance of a single MOS transistor biased by the same current. However, the output resistance of a composite device is larger than that of a single MOS transistor and can be calculated as rds ∼ =rdsD(1 +gmDrdsS). Thus, the intrinsic voltage gain of a self-cascode transistor is larger than that of a single MOS transistor, which further improves the DC voltage gain of the OTA. 56518 VOLUME 11, 2023 M. Kumngern et al.: 31.3 nW, 0.5 V Bulk-Driven OTA for Biosignal Processing FIGURE 2. BPF using OTA, (a) schematic, (b) realization of RMOS, (c) equivalent schematic. The circuit was designed as follows. First, the coefficient m was chosen to be equal to 0.7, in order to increase the voltage gain by around 10 dB with acceptable sensitivity to transistor mismatch. The channel lengths for the MDand MStransistors were chosen to be 2 µm and 1 µm, respectively. Note that relatively large values of L were chosen to increase the rds resistances of the transistors. The nominal biasing current Iset was chosen to be 15 nA. This enables the circuit to obtain the required transconductance value as well as an acceptable value of the input referred thermal noise. The particular values of W for transistors were calculated to obtain dc voltages at an operating point equal to about half-supply for all diode-connected self-cascode transistors. Such a solution increased the available tuning range and provided sufficient headroom for variations of the potentials associated with process, voltage, and temperature (PVT). The relationship between W/L of the MSand MDtransistors was chosen to obtain |VDS|voltages for the MStransistors equal to around 10mV. This provided a good tradeoff between voltage gain and signal swing. Finally, the design was verified by simulations, including noise and Monte Carlo analyses. B. PROPOSED BANDPASS FILTER A conventional BPF is shown in Fig. 2 (a) [44]. The BPF is realized using a single-ended OTA, four capacitors, and two resistors. The capacitor C1eliminates the dc offset voltage from input electrodes. To avoid signal attenuation, the impedance of C1should be much smaller than the impedance of the signal electrode [45]. Utilizing capacitors in the feedback network eliminates the need for resistors with very large values and good linearity, which would be difficult to implement on a chip. The large resistors RMOS in Fig. 2 (a) are used to properly bias the inputs of the OTA for DC. They are realized using two MOS transistors MRwith bulks connected to VDD, as shown in Fig. 2 (b). Because of their very large values, they are bypassed by capacitors for AC. The equivalent schematic of the bandpass filter, used to determine its transfer function, is shown in Fig. 2 (c), where CLrepresents the load capacitance. The voltage gain of the BPF is described by bandpass characteristics. The voltage gain (G) in the middle of its bandwidth can be expressed as: G= C1 C2 1+1+C1 C21 Av (9) where Avis the open-loop voltage gain of OTA, given by (1). Note that for Avtending to infinity, the ideal mid-band voltage gain of the BPF is given by: G=C1 C2(10) Thus, Gis established by the ratio of capacitances only, since in this frequency range their impedances are much lower than RMOS. The lower cutoff frequency of the BPF can be expressed as: fL=1 2πC2RMOS (11) Thus, fLdepends on the time constant being the product of RMOS and C2. Due to the very large value of RMOS, a very low cutoff frequency can be obtained (<1Hz), even for low values of C2, which is desirable in order to limit the value of C1for large voltage gains. The higher cutoff frequency of the BPF can be expressed as: fH= gm1+C1/C2 Av 2πC1 C2(C2+CL)∼ =gm 2πC1 C2(C2+CL) (12) Except for the above poles, the circuit transmittance shows one right half-plane zero located at: fRHP =gm 2πC2(13) The value of fRHP is typically much larger than fH, and most often its impact can be neglected. The input-referred noise of the BPF can be expressed by: V2 ni,amp =C1+C2 C12 ·V2 n(14) where V2 nand V2 ni,amp are the input-referred noise of the OTA and BPF, respectively. Thus, for large voltage gains VOLUME 11, 2023 56519 M. Kumngern et al.: 31.3 nW, 0.5 V Bulk-Driven OTA for Biosignal Processing FIGURE 3. The frequency characteristic of the OTA (a) magnitude (b) phase. FIGURE 4. The magnitude (a) and phase characteristic (b) of the BPF. FIGURE 5. The magnitude (a) and phase characteristic (b) of the BPF for different Iset. (large C1/C2ratios), the input referred noise of the BPF is approximately equal to the input noise of the OTA itself. From the above formulas, one can conclude that the midband gain of the BPF can be controlled by the value of C1 with a constant C2. Further, the lower cutoff frequency can be set by resistor RMOS and the higher cutoff frequency can be adjusted by gm. III. SIMULATION RESULTS The proposed structure was designed in the Cadence platform and simulated with the Spectre simulator of the Analog Design Environment, using a 0.18 µm CMOS process from TSMC where the threshold voltage of the MOS transistor is around 0.5 V. The voltage supply was 0.5 V (±0.25V) and the power consumption was 31.3 nW for 15 nA nominal setting current. The component values of the OTA are given in Table 1. The frequency and phase characteristics of the OTA with a 15 pF load capacitance are shown in Fig. 3 (a) and (b), respectively. The low frequency gain was 54.7 dB, the gain bandwidth was 6.18 kHz and the phase margin was 75◦. The input and output impedance of the OTA at low frequency were 74.7 Gand 890 M, respectively. 56520 VOLUME 11, 2023 M. Kumngern et al.: 31.3 nW, 0.5 V Bulk-Driven OTA for Biosignal Processing FIGURE 6. The magnitude (a) and phase characteristics of the BPF (b) for different C1. TABLE 1. Transistor aspect ratio of the OTA. The magnitude and phase characteristic of the BPF with MR=4µm/15 µm, Iset =15 nA, C1=10 pF, C2=0.1 pF and CL=10 pF are shown in Fig. 4. The mid-band gain was 37.1 dB, the lower and higher cut-off frequency was 1.5 Hz and 112 Hz, respectively. Note that the mid-band gain was slightly lower than the ratio of capacitances C1and C2, due to the limited voltage gain of the OTA (see (9)). The power consumption of the BPF was 31.3 nW. Fig. 5shows the tuning capability of the higher cutoff frequency of the BPF by means of Iset =(5, 15, 25, 35, 45) nA. The fLwas around 1.5 Hz while fH=41.2 Hz, 112 Hz, 186 Hz, 251 Hz and 317 Hz, respectively. Fig. 6 (a) and (b) shows, respectively, the magnitude and phase characteristics of the BPF with Iset =15 nA, C2=0.1 pF and various C1=(1, 2, 5, 10, 20) pF, with CL=10 pF. This was done in order to increase the mid-band gain, which was 18.2 dB, 24.1 dB, 31.7 dB, 37.11 Hz and 41.98 dB, respectively. Based on the transient analysis, Fig. 7shows the spectrum of the output signal using Fast Fourier Transform (FFT) when the input of the BPF is supplied by a sine wave signal with amplitude Vin−amp =1 mV@ 10 Hz and Iset =15 nA. The total harmonic distortion (THD) is obtained based on the FFT that indicates −53.56 dB (0.2%). The THD of the BPF for Vin−amp =1 mV@ 10 Hz versus different Iset =(5, 15, 25, 35, 45) nA and different C1=(1, 2, 5, 10, 20) pF is shown in Fig. 8 (a) and (b), respectively. For all Iset, the THD is below FIGURE 7. The FFT of the output signal. 0.43% and for all C1, the THD is below 1.2%. Thus, the BPF offers low THD. The 200 runs Monte-Carlo (MC) process and mismatch analysis for the magnitude and phase characteristics of the BPF are shown in Fig. 9, indicating acceptable deviation. The histograms of the gain of the BPF @10Hz and the power consumption with Iset =15 nA are shown in Fig. 10 (a) and (b), respectively. The mean value of the gain is 37.1 dB and the standard deviation is 43.6 mdB. The mean value of the power consumption is 31.3 nW with a standard deviation of 492 pW. This confirms the robustness of the design. The process, voltage, and temperature (PVT) corner analysis for the magnitude and phase characteristics of the BPF are shown in Fig 11. The process corners (fast-fast, slowfast, fast-slow, slow-slow), voltage corners (VDD±10%) and temperatures corners (−20◦C, 70◦C) were used. As it is evident, the curves overlap, thus confirming the robustness of the design against PVT corners. The input-referred voltage noise spectrum of the BPF is shown in Fig. 12. The thermal noise level is 0.8 µV√Hz. The integration of the noise from 1.5 Hz to 112 Hz yields a noise 17.9 µVrms. Fig. 13 (a) shows the ECG noisy signal that was applied to the input of the BPF containing low frequency 0.1mV @ 0.1Hz and high frequency 0.1mV @ 2kHz noises. The VOLUME 11, 2023 56521 M. Kumngern et al.: 31.3 nW, 0.5 V Bulk-Driven OTA for Biosignal Processing FIGURE 8. The total harmonic distortion of the output signal with different Iset (a) and different C1(b). FIGURE 9. The magnitude (a) and phase characteristics of the BPF (b) with 200 runs MC. FIGURE 10. The histogram of the gain of the BPF @10Hz (a) and power consumption of the BPF (b) with 200 runs MC. FIGURE 11. The magnitude (a) and phase characteristics (b) with PVT corners. filtered and the amplified output signal is shown in Fig. 13 (b). Table 2shows the performance comparison of the proposed BPF with some previous works [24], [28], [29], [30], [31], 56522 VOLUME 11, 2023 M. Kumngern et al.: 31.3 nW, 0.5 V Bulk-Driven OTA for Biosignal Processing TABLE 2. Comparison of the proposed bpf and some previous works. FIGURE 12. The input-referred voltage noise spectrum. [35], [37], [38]. Compared with BPFs that can control gain and bandwidth in [28] and [29], the proposed circuit clearly has lower power consumption. Compared with BPFs in [24] and [35], the proposed circuit can control gain and has lower power consumption than [24]. Compared with [31] and [38], the proposed BPF can control gain and bandwidth. Although the input referred noise is higher when compared with [28], due to the bulk-driven technique, the proposed circuit has much lower power consumption and voltage supply (31.3 nW, 0.5 V versus 1.1 µW, 1 V). Therefore, the circuit is a tradeoff between the voltage supply and power consumption on one hand and the input referred noise on the other. The THD for the proposed BPF is the lowest compared to other designs. The estimated chip area of the BPF including a 10 pF load capacitance is 0.0167 mm2. Finally, the figure of merit FIGURE 13. The noisy input ECG signal (a) and the filtered output signal (b). (FoM), defined in [37], has been used FoM =P×VDD N×fo×DR (15) VOLUME 11, 2023 56523 M. Kumngern et al.: 31.3 nW, 0.5 V Bulk-Driven OTA for Biosignal Processing where p is the power consumption, VDD is the supply voltage, N is the order, fois the center frequency, DR is the dynamic range. IV. CONCLUSION This paper presents a new BPF using low-voltage and lowpower OTA. The OTA is realized using BD MOST operating in the subthreshold region; thus, it can work with 0.5 V of voltage supply and 31.3 nW of power. The structure was designed in the Cadence program using 0.18 µm CMOS technology from TSMC. The simulated results agree with the theoretical analysis. The proposed BPF can be applied to biomedical signals, such as local field potential, electroencephalogram, and electrocorticography. REFERENCES [1] L. Moldovan and H. H. Li, ‘‘A rail-to-rail, constant gain, buffered op-amp for real time video applications,’’ IEEE J. Solid-State Circuits, vol. 32, no. 2, pp. 169–176, Feb. 1997, doi: 10.1109/4.551908. [2] B. J. Blalock, P. E. Allen, and G. A. Rincon-Mora, ‘‘Designing 1-V op amps using standard digital CMOS technology,’’ IEEE Trans. Circuits Syst. II, Analog Digit. Signal Process., vol. 45, no. 7, pp. 769–780, Jul. 1998, doi: 10.1109/82.700924. [3] B. W. Lee, B. J. Sheu, and H. Yang, ‘‘Analog floating-gate synapses for general-purpose VLSI neural computation,’’ IEEE Trans. Circuits Syst., vol. 38, no. 6, pp. 654–658, Jun. 1991, doi: 10.1109/31.81862. [4] J. Ramirez-Angulo, A. J. Lopez-Martin, R. G. Carvajal, and F. MunozChavero, ‘‘Very low-voltage analog signal processing based on quasi-floating gate transistors,’’ IEEE J. Solid-State Circuits, vol. 39, no. 3, pp. 434–442, Mar. 2004, doi: 10.1109/JSSC.2003.822782. [5] F. Assaderaghi, D. Sinitsky, S. A. Parke, J. Bokor, P. K. Ko, and C. Hu, ‘‘Dynamic threshold-voltage MOSFET (DTMOS) for ultra-low voltage VLSI,’’ IEEE Trans. Electron Devices, vol. 44, no. 3, pp. 414–422, Mar. 1997, doi: 10.1109/16.556151. [6] J. M. Carrillo, G. Torelli, R. Prez-Aloe, and J. F. Duque-Carrillo, ‘‘1-V rail-to-rail CMOS OpAmp with improved bulk-driven input stage,’’ IEEE J. Solid-State Circuits, vol. 42, no. 3, pp. 508–517, Mar. 2007, doi: 10.1109/JSSC.2006.891717. [7] L. Zuo and S. K. Islam, ‘‘Low-voltage bulk-driven operational amplifier with improved transconductance,’’ IEEE Trans. Circuits Syst. I, Reg. Papers, vol. 60, no. 8, pp. 2084–2091, Aug. 2013, doi: 10.1109/TCSI.2013.2239161. [8] J. Ramirez-Angulo, R. G. Carvajal, J. Tombs, and A. Torralba, ‘‘Lowvoltage CMOS op-amp with rail-to-rail input and output signal swing for continuous-time signal processing using multiple-input floating-gate transistors,’’ IEEE Trans. Circuits Syst. II, Analog Digit. Signal Process., vol. 48, no. 1, pp. 111–116, Jan. 2001, doi: 10.1109/82.913195. [9] T. Thongleam, A. Suadet, and V. Kasemsuwan, ‘‘A 0.8 V quasi-floatinggate fully differential CMOS op-amp with positive feedback,’’ in Proc. 8th Electr. Eng./Electron., Comput., Telecommun. Inf. Technol. (ECTI), May 2011, pp. 98–101, doi: 10.1109/ECTICON.2011.5947780. [10] S. Dong, Y. Wang, X. Tong, Y. Wang, and C. Liu, ‘‘A 0.3-V 8.72-nW OTA with bulk-driven low-impedance compensation for ultra-low power applications,’’ Circuits, Syst., Signal Process., vol. 40, no. 5, pp. 2209–2227, May 2021, doi: 10.1007/s00034-020-01590-9. [11] T. Kulej, F. Khateb, D. Arbet, and V. Stopjakova, ‘‘A 0.3-V high linear rail-to-rail bulk-driven OTA in 0.13 µm CMOS,’’ IEEE Trans. Circuits Syst. II, Exp. Briefs, vol. 69, no. 4, pp. 2046–2050, Apr. 2022, doi: 10.1109/TCSII.2022.3144095. [12] B. Aggarwal and A. Gupta, ‘‘QFGMOS and FGMOS based lowvoltage high performance MI-OTA,’’ Int. J. Inf. Technol., vol. 13, no. 2, pp. 415–422, Apr. 2021, doi: 10.1007/s41870-020-00541-6. [13] S. A. Katourani, O. Hashemipour, and M. Meghdadi, ‘‘An ultra high slew rate recycling folded cascode OTA with 100-dB DC gain,’’ AEUInt. J. Electron. Commun., vol. 147, Apr. 2022, Art. no. 154140, doi: 10.1016/j.aeue.2022.154140. [14] M. Mahendra, S. Kumari, and M. Gupta, ‘‘Low voltage fully differential OTA using DTMOS based self cascode transistor with slew-rate enhancement and its filter application,’’ Integration, vol. 84, pp. 47–61, May 2022, doi: 10.1016/j.vlsi.2022.01.003. [15] F. Khateb and T. Kulej, ‘‘Design and implementation of a 0.3-V differential difference amplifier,’’ IEEE Trans. Circuits Syst. I, Reg. Papers, vol. 66, no. 2, pp. 513–523, Feb. 2019, doi: 10.1109/TCSI.2018.2866179. [16] F. Khateb, M. Kumngern, S. Vlassis, C. Psychalinos, and T. Kulej, ‘‘Sub-volt fully balanced differential difference amplifier,’’ J. Circuits, Syst. Comput., vol. 24, no. 1, Jan. 2015, Art. no. 1550005, doi: 10.1142/S021812661550005X. [17] F. Khateb, S. Vlassis, M. Kumngern, C. Psychalinos, T. Kulej, R. Vrba, and L. Fujcik, ‘‘1 V rectifier based on bulk-driven quasi-floating-gate differential difference amplifiers,’’ Circuits, Syst., Signal Process., vol. 34, no. 7, pp. 2077–2089, Jul. 2015, doi: 10.1007/s00034-014-9958-3. [18] T. Kulej and F. Khateb, ‘‘0.4-V bulk-driven differential-difference amplifier,’’ Microelectron. J., vol. 46, no. 5, pp. 362–369, May 2015, doi: 10.1016/j.mejo.2015.02.009. [19] F. Khateb, T. Kulej, and M. Kumngern, ‘‘0.3 V bulk-driven current conveyor,’’ IEEE Access, vol. 7, pp. 65122–65128, 2019, doi: 10.1109/ACCESS.2019.2916897. [20] F. Khateb, T. Kulej, M. Kumngern, W. Jaikla, and R. K. Ranjan, ‘‘Comparative performance study of multiple-input bulk-driven and multipleinput bulk-driven quasi-floating-gate DDCCs,’’ AEU-Int. J. Electron. Commun., vol. 108, pp. 19–28, Aug. 2019, doi: 10.1016/j.aeue.2019. 06.003. [21] F. Khateb, N. Khatib, and D. Kubanek, ‘‘Low-voltage ultra-low-power current conveyor based on quasi-floating gate transistors,’’ Radioengineering, vol. 2, pp. 725–735, Jun. 2012. [22] A. Uygur and H. Kuntman, ‘‘A very compact, 0.4 V DTMOS CCII employed in an audio-frequency filter,’’ Anal. Integr. Circuits Signal Process., vol. 81, no. 1, pp. 89–98, Oct. 2014, doi: 10.1007/s10470-014-03652. [23] C. Sun and S. Lee, ‘‘A fifth-order Butterworth OTA-C LPF with multipleoutput differential-input OTA for ECG applications,’’ IEEE Trans. Circuits Syst. II, Exp. Briefs, vol. 65, no. 4, pp. 421–425, Apr. 2018, doi: 10.1109/TCSII.2017.2695366. [24] F. Khateb, P. Prommee, and T. Kulej, ‘‘MIOTA-based filters for noise and motion artifact reductions in biosignal acquisition,’’ IEEE Access, vol. 10, pp. 14325–14338, 2022, doi: 10.1109/ACCESS.2022. 3147665. [25] S. Lee, C. Wang, and Y. Chu, ‘‘Low-voltage OTA-C filter with an areaand power-efficient OTA for biosignal sensor applications,’’ IEEE Trans. Biomed. Circuits Syst., vol. 13, no. 1, pp. 56–67, Feb. 2019, doi: 10.1109/TBCAS.2018.2882521. [26] M. Kumngern, F. Khateb, T. Kulej, D. Arbet, and M. Akbari, ‘‘Fully differential fifth-order dual-notch low-pass filter for portable EEG system,’’ AEU-Int. J. Electron. Commun., vol. 146, Mar. 2022, Art. no. 154122, doi: 10.1016/j.aeue.2022.154122. [27] M. Kumngern, N. Aupithak, F. Khateb, and T. Kulej, ‘‘0.5 V fifthorder Butterworth low-pass filter using multiple-input OTA for ECG applications,’’ Sensors, vol. 20, no. 24, p. 7343, Dec. 2020, doi: 10.3390/s20247343. [28] P. Sharma, K. Sharma, H. S. Jatana, J. Madan, R. Pandey, and R. Sharma, ‘‘A 1.1 µW biopotential amplifier based on bulk-driven quasi-floating gate technique with extremely low-value of offset voltage,’’ Anal. Integr. Circuits Signal Process., vol. 103, no. 2, pp. 303–313, May 2020. [Online]. Available: https://link.springer.com/article/10.1007/s10470-020-01623-y [29] F. Khateb, T. Kulej, M. Kumngern, D. Arbet, and W. Jaikla, ‘‘A 0.5V 95-dB rail-to-rail DDA for biosignal processing,’’ AEU-Int. J. Electron. Commun., vol. 145, Feb. 2022, Art. no. 154098, doi: 10.1016/j.aeue.2021.154098. [30] C. D. Salthouse and R. Sarpeshkar, ‘‘A practical micropower programmable bandpass filter for use in bionic ears,’’ IEEE J. Solid-State Circuits, vol. 38, no. 1, pp. 63–70, Jan. 2003. [31] I. Corbacho, J. M. Carrillo, J. L. Ausín, M. Á. Domínguez, R. Pérez-Aloe, and J. F. Duque-Carrillo, ‘‘CMOS widely tunable second-order Gm-C bandpass filter for multi-sine bioimpedance analysis,’’ Electronics, vol. 12, no. 6, p. 1326, Mar. 2023, doi: 10.3390/electronics12061326. [32] J. M. Carrillo and C. A. de la Cruz-Blas, ‘‘0.6-V 1.65-µW second-order Gm-C bandpass filter for multi-frequency bioimpedance analysis based on a bootstrapped bulk-driven voltage buffer,’’ J. Low Power Electron. Appl., vol. 12, no. 4, p. 62, Nov. 2022, doi: 10.3390/jlpea12040062. [33] Y. Sundarasaradula and A. Thanachayanont, ‘‘A 1-V, 6-nW programmable 4th-order bandpass filter for biomedical applications,’’ Anal. Integr. Circuits Signal Process., vol. 89, no. 1, pp. 89–98, Oct. 2016, doi: 10.1007/s10470-016-0792-3. 56524 VOLUME 11, 2023