- Email: [email protected]
A novel low-power transceiver topology for noncontact vital sign detection including the power management technique
Microelectronics Journal 44 (2013) 1309–1315
Contents lists available at ScienceDirect
Microelectronics Journal journal homepage: www.elsevier.com/locate/mejo
A novel low-power transceiver topology for noncontact vital sign detection including the power management technique Chie-In Lee a,b,n, Yan-Ting Lin a, Yu-Her Chen a, Wei-Cheng Lin a a b
Department of Electrical Engineering, National Sun Yat-Sen University, No. 70 Lienhai Rd., Kaohsiung 80424, Taiwan Institute of Communications Engineering, National Sun Yat-Sen University, No. 70 Lienhai Rd., Kaohsiung 80424, Taiwan
art ic l e i nf o
a b s t r a c t
Article history: Received 19 April 2013 Received in revised form 28 August 2013 Accepted 29 August 2013 Available online 28 October 2013
In this paper, power management technique utilized in the direct down-conversion non-quadrature transceiver is presented for the low-power application of vital sign detection. The simultaneous switching noise (SSN) and overshoot and undershoot of the transient waveform distortion resulting from a pulse signal will give rise to interference with a vital sign signal. The pulse width, rise/fall time, and period of pulse bias are analyzed to mitigate the interference in this investigation. Significant issues about direct-current (DC) offset and noise confronted by the presented technique are addressed based on mathematical analysis. In radio-frequency (RF) transceiver architecture including power amplifier (PA), low-noise amplifier (LNA), and mixer, the current-reused (CRU) topology is utilized to achieve low DC power consumption. The post-layout simulation results exhibit that power consumption of the transceiver using the optimized pulse bias is reduced to 40% of the power consumption for transceiver applying the DC bias. In addition, DC offset and null detection point can be alleviated by tunable phase shifter. Crown Copyright & 2013 Published by Elsevier Ltd. All rights reserved.
Keywords: Vital sign detection Power management technique Switching noise Power amplifier
1. Introduction The noncontact Doppler radar sensor requires ultra low-power circuit and high accuracy to detect the vital sign signals for longterm health care [1–3] and life detection [4]. To improve detection accuracy, many topologies have been presented. The quadrature architecture designed for a wireless communication system [5,6] is utilized to detect the vital sign with channel selection [1,2]. In addition, the double-sideband transmission system with wide frequency tuning range [3] is presented to eliminate the null detection point. However, these structures require several active components so that the direct-current (DC) power consumption of the whole system increases [7]. The direct down-conversion nonquadrature topology can be implemented in the low-power vital sign detection system due to less active components required. In the front-end circuit design, several circuit architectures have been developed for low-power application [8–10]. Complementary structure has been presented [8] at the expense of a degraded power gain to provide low-voltage capability. To enhance the gain in low-power design, the current-reused (CRU) configuration [9,10] is proposed to merge the currents required for two transistors into a single current path such that supply current can be reduced. Power consumption is only half of the cascade topology, n Corresponding author at: Department of Electrical Engineering, National Sun Yat-Sen University, No. 70 Lienhai Rd., Kaohsiung, Taiwan 80424. Tel.: þ 886 7 525 2000 4110; fax: þ886 7 525 4199. E-mail address: [email protected] (C.-I. Lee).
and gain can be approximately the same with the result of the cascade topology. From another point of view, the power management technique can be employed to further improve efficiency [11–13]. The buck–boost converter can provide dynamic power supply to the power amplifier (PA) so as to prolong the battery life. However, the switching voltage may produce simultaneous switching noise (SSN) [14,15] as well as overshoot and undershoot. These voltage ripples should be mitigated to acceptable level in order to avoid interference with the vital sign signals. Our previous research presented a high linearity and lowpower low-noise amplifier (LNA) design with nonlinear cancellation technique [16]. In this paper, a low-power direct downconversion non-quadrature transceiver with the power management technique that is applied for the noncontact vital sign detection is presented for the first time. SSN and transient waveform distortion are analyzed and minimized to improve detection accuracy of the presented system with pulse bias. For the transceiver design, the CRU topology is employed to further reduce DC power consumption of the whole system.
2. Power management technique and vital sign detection theory The main subject of this work is to design a low-power radiofrequency (RF) transceiver for the application of vital sign detection and to maintain detection accuracy simultaneously. Fig. 1 shows novel direct down-conversion non-quadrature transceiver
0026-2692/$ - see front matter Crown Copyright & 2013 Published by Elsevier Ltd. All rights reserved. http://dx.doi.org/10.1016/j.mejo.2013.08.021
1310
C.-I. Lee et al. / Microelectronics Journal 44 (2013) 1309–1315
Fig. 1. The novel direct down-conversion non-quadrature transceiver with the power management technique and the tunable phase shifter.
with the power management technique. This is the first time that the power management technique with pulse bias is implemented in the presented PA to further decrease power consumption and to alleviate interference of the SSN, overshoot, and undershoot. Fig. 2 describes the design procedure for the presented vital sign sensor with SSN and overshoot/undershoot considered. In the active state of pulse bias, maximum output power of the PA is 9.3 dBm as the drain voltage is operated at high voltage (VH) of 2.6 V. For the sleep state, the minimum output power is 7 dBm at low voltage (VL) of 0.6 V. When the pulse signal is applied to the circuits, variation of the pulse bias will produce signal reflection at the boundary due to transmission line discontinuities [17]. The reflection leads to instantaneous change of the current and voltage at each node, causing overshoot and undershoot and hence significant ripple in the transmitted signal as shown in Fig. 3(a). Besides, SSN can be generated in power and ground planes due to the rapid change in current flowing through parasitic inductor of the power supply [14,18]. Therefore, to investigate the influence of the pulse signal on the presented system, the microstrip lines, the circuit layout, and the bonding wires of the presented system are considered in the simulation. On the other hand, a DC offset effect can occur in direct down-conversion non-quadrature architecture when a RF signal at the same frequency with the local oscillator (LO) frequency is mixed with the LO signal. Significant issues about DC offset and noise confronted by the presented technique will be addressed based on mathematical analysis. The transmitted signal of a transmitter influenced by transient waveform distortion due to pulse bias can be expressed as follows: TðtÞ ¼ AðtÞ sin ð2πf t þ φÞ
ð1Þ
where T(t) is the transmitted signal from the transmitter, f is the carrier frequency, and φ is the residual phase that is accumulated in the system. The ripple voltage of the transmitted signal due to transient waveform distortion is given by A(t). When this transmitted signal is reflected by a human subject at a nominal distance d0, the periodic body movements x(t) due to human cardiopulmonary activity will be modulated in the phase of the received signal according to the Doppler theory [2]. In addition, voltage amplitude variation still exists in the received signal along with the vital sign signal. Therefore, the received signal R(t) can be expressed as 4π 4πd0 RðtÞ ¼ Rr ðtÞ sin 2πf t xðtÞ þφ ð2Þ λ λ where Rr(t) is the amplitude of the received signal with the timevarying ripple voltage and λ is the wavelength of the RF signal. Information about the human cardiopulmonary activity can be demodulated through the mixer with an LO signal, which is derived from the transmitted signal. Note that the phase noise from LO can
Fig. 2. The flow chart describes the design process for the power management technique based on the vital sign detection.
be neglected due to range correlation [2,19]. Therefore, a detected signal in the baseband B(t) can be expressed as BðtÞ ¼ Br ðtÞ cos ðφB
4πxðtÞ Þ λ
ð3Þ
where φB ¼ φ
4πd0 φLO λ
ð4Þ
Br(t) is the amplitude of the baseband signal with the timevarying ripple voltage and φLO is the phase of the LO signal. Based on the Taylor series expansion, the baseband signal shown in Eq. (3) can be approximated as BðtÞ ¼ Br ðtÞ sin φB U
4π U xðtÞ þ Br ðtÞ cos φB λ
ð5Þ
High-order terms of the expansion are neglected under the condition xðtÞ o o λ. The first term on the right-hand side of Eq. (5) contains the vital sign signal as well as the ripple voltage at baseband Br(t). This ripple voltage can be a significant noise source, which can overwhelm the vital sign signal, degrading the detection accuracy. The second term on the right-hand side of Eq. (5)
C.-I. Lee et al. / Microelectronics Journal 44 (2013) 1309–1315
1311
Fig. 3. (a) Schematic diagram of the presented class-E PA with a stacked DPTF and a balun. Blue lines show the overshoot and undershoot of the transient waveform distortion occurred in the circuit when the pulse bias at the drain terminal of the PA is employed. (b) Layout of the class-E CRU PA. (c) Schematic diagram of the CRU LNA with a differential inductor. (d) Layout of the CRU LNA. (e) Schematic diagram of the CRU mixer. (f) Layout of the CRU mixer. (For interpretation of the references to color in this figure legend, the reader is referred to the web version of this article.)
represents the DC offset. This DC term has to be considered especially for the direct down-conversion non-quadrature architecture. In noncontact vital sign detection, the DC offset usually comes from clutter reflection and from circuit imperfection. Because the DC offset due to circuit imperfection is alleviated by high RF–LO isolation mixer [20], clutter reflection becomes the dominant factor in the Doppler vital sign detection [20]. Therefore, the source of DC offset considered in this paper is from clutter reflection. In Eq. (5), in addition to the time-varying movement of the vital sign, undesired time-varying ripple amplitude Br(t) is also included at baseband. If ripple voltage of the transmitted signal is large, detection accuracy will be significantly decreased. As a result, appropriate pulse rise/fall time is selected to avoid transient waveform distortion and to minimize amplitude variation for vital sign detection. Furthermore, to achieve accurate detection, pulse period is designed as short as possible such that
the vital sign signals can almost occur in the active state. Nevertheless, SSN will increase when a short period or rise/fall time pulse bias is adopted [14]. Therefore, as a compromise between the accurate detection and switching noise, the period and rise/fall time of the pulse bias are optimized for vital sign detection. As the pulse period and rise/fall time are fixed, the pulse width can be shortened to reduce power consumption for the presented system. On the other hand, the transmitted power is proportional to the pulse width. Therefore, selection of the pulse width is a tradeoff between overall power consumption and detection accuracy. To maintain vital sign detection accuracy with low power consumption, the pulse width is shortened such that amplitude of the vital sign signals in the baseband channel can still be 10–20 dB larger than the noise floor required for noncontact vital sign detection [19].
1312
C.-I. Lee et al. / Microelectronics Journal 44 (2013) 1309–1315
3. The transceiver design In the transmitter architecture of the presented system, the CRU topology is utilized to reduce power consumption of the PA. The output power of CRU PA is improved by using the transformer to combine the power delivered from several power cells. Fig. 3 (a) shows the architecture of a fully integrated PA operated in class-E mode consisting of the balun, the stacked double primary transformer (DPTF), and two push-pull PAs. The transistors (M1–M4) are in common-source configuration. An inter-stage network consisting of coupling capacitances C1 and C2 is designed for better signal matching. DC current flows through stacked transistors due to the low reactance of inductances (L1, L2) as shown in Fig. 3(a). Amplified RF signals from M1 and M2 are impeded by L1 and L2, and thus RF signals are fed into the gate terminal of M3 and M4 through C1 and C2. Therefore, the overall transconductance of this power stage can be equal to the value of the cascade topology. The self-bias topology is utilized in the driver stages to prevent the degradation of CMOS devices. Table 1 summarizes the post-layout simulated PA performance using CMOS 0.18 μm technology design kit provided by Taiwan Semiconductor Manufacturing Company (TSMC), Hsinchu, Taiwan. The power consumption of CRU PA is decreased to about 62.3% of conventional cascode PA, which is reduced from 47 mW to 29.3 mW, and the output power is only slightly degraded. To further enhance the output power under the condition of low power consumption, passive balun and DPTF are adopted for input and output matching of the presented PA, respectively. Input balun transforms the input single-ended signal to the differential signal as shown in Fig. 3(a). The output ports of the driver stages are connected to input ports of the power stages through capacitance and to one of the two primary coil of DPTF in order to improve the output power. In addition, the output power from power stages is fed to the other primary coil. The presented DPTF structure consisting of two three-turn primary coils and a single three-turn secondary coil is employed to enhance the output power, and current directions of the primary coils are the same to avoid self-cancellation. The insertion loss of DPTF from the silicon substrate is mitigated by connecting output port of the DPTF with an additional metal–insulator–metal capacitance. The layout of the presented class-E CRU PA with passive components is shown in Fig. 3(b). The output power of the whole PA with passive components is higher than the values of CRU PA and conventional cascode PA as shown in Table 1. Transmitted power at fundamental frequency of 2.4 GHz from the presented PA is about 9.3 dBm, with the DC power consumption of 28.1 mW. Therefore, this CRU PA with the balun and DPTF can achieve high output power and low power consumption simultaneously for remote vital sign detection or life detection of buried victims under rubble [4]. For the receiver architecture, a complementary CRU LNA and a CRU mixer are designed to achieve low power consumption. The complementary CRU LNA consists of three stages in commonsource configuration as shown in Fig. 3(c). The first stage of the CRU LNA includes a differential inductor connected at the source terminal of M1. The phase of the gate and source voltage of the M1 transistor can be opposite by using the differential inductor so
Table 1 The output power and the power consumption for different PA topologies. Topology
Conventional cascode PA
CRU PA
CRU PA with the DPTF and balun
Maximum Output power (dBm) Power consumption(mW)
4.4 47
3.2 29.3
9.3 28.1
that the transconductance of input transistor can be increased. Therefore, gain of the whole LNA circuit increases larger than the conventional CRU topology under the condition of low power consumption. In addition, the differential inductor provides inductive compensation for an input network to reduce the noise figure. The second and third stages consisting of M2 and M3, respectively, utilize a complementary topology to further reduce the DC supply voltage. L1, C1, and C2 are employed for inter-stage matching. The layout of complementary CRU LNA is shown in Fig. 3(d). Postlayout simulated gain and noise figure at 2.4 GHz are 10.1 dB and 3.4 dB, respectively. It is known that the performance of the direct down-conversion receiver may suffer from the flicker noise [21]. Therefore, in the mixer design, a CRU topology consisting of M1 and M2 transistors shown in Fig. 3(e) is utilized. With the CRU topology, current flowing through the switch stage M2 can be reduced so that flicker noise of the mixer can be mitigated [21,22]. The layout of CRU mixer is shown in Fig. 3(f). Post-layout simulation shows 5 dB and 7 dB of gain and noise figure, respectively. In receiver architecture, tunable phase shifter shown in Fig. 1 is employed to alleviate the DC offset and null detection point [23,24]. Sensitivity is a significant parameter for the receiver. To estimate sensitivity of the presented system, receiver sensitivity is determined based on the formula given as [19] Sensitivity ðdBmÞ ¼ kTB ðdBmÞ þNF r ðdBÞ þ SNR ðdBÞ
ð6Þ
where T is the temperature in kelvin, B is the receiver output bandwidth, k is Boltzmann's constant, NFr is the noise figure in the receiving path, and SNR is the signal-to-noise ratio required for the noncontact vital sign detection, which is 10–20 dB typically [19]. Assuming that the phase noise from the LO is neglected due to the range correlation [2,19], the sensitivity of the presented sensor is calculated to be 82.5 dBm using Eq. (6) at room temperature 290 K, with a bandwidth of 1 MHz [19], noise figure of 16.5 dB in the receiving path, and SNR of 15 dB within the typical range of 10–20 dB [19]. The Advanced Technical Materials 340-441-2 horn antennas with 15 dBi gain are used to transmit a 2.4 GHz signal with maximum output power of 9.3 dBm and to receive the reflected signal from the human body. The parameters of the components mentioned above are utilized in the simulator. Maximum detecting distance between the antenna and the human body can be 4.4 m approximately based on the radar equation [19] with a typical radar cross section of the human heartbeat at 2.4 GHz [25]. Therefore, the presented sensor can detect vital sign signals within 4.4 m.
4. Results and discussion Power management technique is applied to the presented vital sign detection system to further decrease power consumption. To determine the optimal pulse bias for satisfying the requirement of SSN and waveform distortion with low power consumption, output characteristics of the sensor with different pulse periods, rise/ fall time Tr,f, and width are simulated. Acceptable SSN level is required to be lower than 10% of the output voltage amplitude [26]. As for the waveform distortion, overshoot and undershoot currents have to be below 5% of the output current amplitude [27]. In Fig. 4(a), voltage fluctuation of the pulse biased sensor due to SSN decreases with increase in pulse rise/fall time. It is also observed that voltage fluctuation is almost independent of the pulse period as the pulse period is longer than 1 ms as shown in the inset of Fig. 4(a). Therefore, optimal pulse period is chosen as 1 ms as a compromise between SSN interference and accurate detection. As for the rise/fall time, the value of 60 ns is selected to satisfy the acceptable level for both SSN and overshoot/undershoot
C.-I. Lee et al. / Microelectronics Journal 44 (2013) 1309–1315
1313
Fig. 4. (a) The SSN versus rise/fall time at different pulse periods of 0.1 ms (dash–dot), 0.4 ms (dot), 0.7 ms (dash), and 1.0 ms (solid). In the inset, the SSN with the pulse periods of 1.0 ms (solid), 10 ms (dash–dot–dot), 20 ms (short dash), and 30 ms (short dot) are shown. (b) The overshoot current (circles), undershoot (squares), and output ripple voltage (triangles) versus rise/fall time at the pulse bias with VH/VL ¼ 2.6/0.6 V and 1 ms pulse period.
Fig. 5. (a) The detected signal spectrum with the different ripple voltages is shown. The optimal rise/fall time of 60 ns corresponds to the ripple voltage of 0.085 V. To purely analyze the influence of rise/fall time on the signal spectrum, the pulse width approximates to one period. (b) The vital sign signals and the DC offset versus the LO phase under the DC bias condition.
currents as shown in Fig. 4(a and b). In Fig. 5(a), we further investigate the effect of the ripple voltage resulting from overshoot/undershoot on vital sign signal spectrum at baseband due to the pulse bias applied. Noise floor increases with the increase in ripple voltage because amplitude variation due to the ripple significantly affects the baseband signals based on Eq. (5). The large amplitude of the respiration signal is not influenced by the ripple voltage. However, the significant noise floor due to large ripple voltage overwhelms the heartbeat signal as shown in Fig. 5 (a). If the ripple voltage is reduced to 0.085 V corresponding to optimal rise/fall time of 60 ns for the overshoot/undershoot requirement shown in Fig. 4(b), the accurate heartbeat signal can be obtained as shown in Fig. 5(a). This indicates that the requirement of overshoot/undershoot current can prevent the baseband signal from being influenced by noise resulting from the timevarying ripple voltage, and thus ensure reliable vital sign detection. Therefore, the optimal pulse period and rise/fall time applied to the presented CRU PA are designed to be 1 ms and 60 ns, respectively. To prevent the vital sign signal from being influenced by the DC offset and null detection point, LO phase is tuned through the tunable phase shifter. The DC offset can be reduced to 0.32 mV when the LO phase is 701, and the maximum amplitude of the vital sign signals can be achieved without the null detection point influenced as shown in Fig. 5(b). In Fig. 6(a–c), maximum vital sign signals and the minimum DC offset can be observed in different pulse width as LO phase is 701, which is consistent with the value obtained from the result with DC bias as shown in Fig. 5(b). Therefore, this demonstrates that the
phase shifter can still provide the function of LO phase tuning to increase detection accuracy even when the pulse bias is applied in the presented system. Fig. 6(b–d) exhibits a tradeoff between power consumption and the detected accuracy. When pulse width is larger than or equal to 0.3 ms, both the amplitudes of respiration and heartbeat signals fulfill the required 10–20 dB signal-to-noise ratio for the noncontact vital sign detection [19]. Therefore, to save DC power consumption without sacrificing the accuracy, optimal pulse width is selected to be 0.3 ms. In summary, optimal pulse bias condition is 1 ms period, 60 ns rise/fall time, and 0.3 ms width. This pulse bias can be provided by the commercial DC–DC converter module. To attain the performances of the sensor purely with pulse bias, the presented system without the usage of DC–DC converter is employed in the simulation environment. In Fig. 7, vital sign signals including the respiration and heartbeat signals are determined by using the presented system with optimal LO phase and pulse period, rise/fall time, and width. Therefore, power management technique not only reduces power consumption of the system but also maintains the capability of vital sign detection. Table 2 summarizes the performances of the low-power systems based on simulation. The power consumption of RF receiver and CRU PA in the presented system by using the optimal pulse bias is 6.6 mW and 8.4 mW, respectively. Therefore, low power consumption is achieved in the presented topology. Other characteristics are in acceptable ranges when compared with performances from other works [28–30]. Based on the above analysis, low power and high accuracy can be simultaneously achieved in the presented system using optimal pulse bias.
1314
C.-I. Lee et al. / Microelectronics Journal 44 (2013) 1309–1315
Fig. 6. Analysis of (a) the DC offset versus the LO phase and the pulse width, (b) the respiration versus the LO phase and the pulse width, (c) the heartbeat versus the LO phase and the pulse width, and (d) the power consumption of the PA versus the pulse width.
Table 2 The performance summary of the low power system for the sensor applications. Specification
This issue
[28]
[29]
[30]
Technology Mode
0.18 μm CMOS TX
RX
0.13 μm CMOS RX
0.18 μm CMOS TX RX
0.35 μm CMOS TX RX
TX power (dBm) Bandwidth (MHz) Noise figure (dB) Power consumption (mW)
9.3 200 4.1a 8.4/21.4c
6.6
N/A 15 8.5a 7.2
N/A 7500 3b 40
0 7 10 30
40
a
The noise figure of the cascaded system (LNA þmixer). The noise figure of the LNA. The power consumption of the proposed transceiver includes 13 mW of the AIC2354 DC/DC converter for practical purposes. b c
Fig. 7. At the LO phase of 701, the detected signal spectrum by using the optimal pulse bias. (VH/VL ¼2.6/0.6 V, Tr,f ¼60 ns, the pulse period ¼1 ms, and the pulse width ¼ 0.3 ms)
5. Conclusion In this paper, power management technique is applied to 2.4 GHz direct down-conversion non-quadrature transceiver to reduce power consumption for noncontact vital sign detection for the first time. The DC offset and noise correlated to the presented system have been
addressed by in-depth analysis. Optimal pulse period, width, and rise/fall time are determined to be 1 ms, 0.3 ms, and 60 ns, respectively. By using this optimal pulse bias, SSN can meet the requirement and the overshoot and undershoot of current can be reduced to 5% of the output current amplitude so that vital sign detection does not interfere with output voltage ripples. The CRU topology is utilized in the transceiver to achieve low power consumption. The total power consumption of the presented system using the optimal pulse
C.-I. Lee et al. / Microelectronics Journal 44 (2013) 1309–1315
bias is approximately 60% less than the result of the system using the DC bias. Furthermore, the tunable phase shifter is used to alleviate the null detection point and DC offset. Therefore, the accurate vital sign detection under low-power operating condition can be achieved by using the presented power management technique. Acknowledgments The authors would like to thank National Chip Implementation Center (CIC), Hsinchu, Taiwan for TSMC design kits, and the Wireless Communication Antenna Research Center, Kaohsiung, Taiwan for the support. This work is also supported in part by the National Science Council of Taiwan under Grant NSC101-2221E-110-077 and under Grant NSC100-2221-E-110-085. References [1] B.K. Park, O. Boric-Lubecke, V.M. Lubecke, Arctangent demodulation with DC offset compensation in quadrature doppler radar receiver systems, IEEE Transactions on Microwave Theory and Techniques 55 (5) (2007) 1073–1079. [2] A.D. Droitcour, O. Boric-Lubecke, V.M. Lubecke, J. Lin, G.T.A. Kovacs, Range correlation and I/Q performance benefits in single-chip silicon Doppler radars for noncontact cardiopulmonary monitoring, IEEE Transactions on Microwave Theory and Techniques 52 (3) (2004) 838–848. [3] Y. Xiao, J. Lin, O. Boric-Lubecke, V.M. Lubecke, Frequency–tuning technique for remote detection of heartbeat and respiration using low-power double-sideband transmission in the Ka-band, IEEE Transactions on Microwave Theory and Techniques 54 (5) (2006) 2023–2032. [4] J.B.H. Tahar, J.C. Bolomey, M. Cauterman, Microwave life detector for buried victims, in: Proceedings of the 23rd European Microwave Conference, 1993, pp. 263–265. [5] Q. Wan, C. Wang, A low-voltage low-power CMOS transmitter front-end using current mode approach for 2.4 GHz wireless communications, Microelectronics Journal 42 (5) (2011) 766–771. [6] M.A. Abdelghany, R.K. Pokharel, H. Kanaya, K. Yoshida, A low flicker noise direct conversion receiver for IEEE 802.11 g wireless LAN using differential active inductor, Microelectronics Journal 42 (2) (2011) 283–290. [7] C. Li, Y. Xiao, J. Lin, Design guidelines for radio frequency non-contact vital sign detection, in: Proceedings of the 29th Annual International IEEE EMBS Conference, 2007, pp. 1651–1654. [8] F. Roewer, U. Kleine, A novel class of complementary folded-cascode opamps for low voltage, IEEE Journal of Solid-State Circuits 37 (8) (2002) 1080–1083. [9] L. Zheng, H.C. Yao, F. Tzeng, P. Heydari, Design and analysis of a current-reuse transmitter for ultra-low power applications, in: Proceedings of the International Symposium on Circuits Systems, 2009, pp. 1317–1320. [10] H.-H. Hsieh, L.-H. Lu, Design of ultra-low-voltage RF frontends with complementary current-reused architectures, IEEE Transactions on Microwave Theory and Techniques 55 (7) (2007) 1445–1458. [11] B. Sahu, G.A. Rincon-Mora, A high-efficiency linear RF power amplifier with a power-tracking dynamically adaptive buck-boost supply, IEEE Transactions on Microwave Theory and Techniques 52 (1) (2004) 112–120.
1315
[12] F.H. Raab, B.E. Sigmon, R.G. Myers, R.M. Jackson, L-band transmitter using Kahn EER technique, IEEE Transactions on Microwave Theory and Techniques 46 (12) (1998) 2220–2225. [13] M. Iwamoto, A. Williams, P.-F. Chen, A.G. Metzger, L.E. Larson, P.M. Asbeck, An extended Doherty amplifier with high efficiency over a wide power range, IEEE Transactions on Microwave Theory and Techniques 49 (12) (2001) 2472–2479. [14] M.A. Méndez, J.L. González, D. Mateo, A. Rubio, An investigation on the relation between digital circuitry characteristics and power supply noise spectrum in mixed-signal CMOS integrated circuits, Microelectronics Journal 36 (1) (2005) 77–84. [15] K.B. Unchwaniwala, M.F. Caggiano, R.C. Frye, Noise generation, propagation and effects on RF components in system-on-a-chip packaging, Microelectronics Journal 33 (5-6) (2002) 471–478. [16] C.I. Lee, W.C. Lin, J.M. Lin, Low-power and high-linearity SiGe HBT low-noise amplifier using IM3 cancellation technique, Microelectronics Engineering 91 (2012) 59–63. [17] R. Achar, M.S. Nakhla, Simulation of high-speed interconnects, Proceedings of the IEEE 89 (5) (2001) 693–728. [18] A. Kabbani, A.J. Al-Khalili, Estimation of ground bounce effects on CMOS circuits, IEEE Transactions on Components, Packaging and Manufacturing Technology 22 (2) (1999) 316–325. [19] C. Li, X. Yu, C.-M. Lee, D. Li, L. Ran, J. Lin, High-sensitivity software-configurable 5.8-GHz radar sensor receiver chip in 0.13-μm CMOS for noncontact vital sign detection, IEEE Transactions on Microwave Theory and Techniques 58 (5) (2010) 1410–1419. [20] C. Gu, R. Li, H. Zhang, A.Y.C. Fung, C. Torres, S.B. Jiang, C. Li, Accurate respiration measurement using DC-coupled continuous-wave radar sensor for motionadaptive cancer radiotherapy, IEEE Transactions on Biomedical Engineering 59 (11) (2012) 3117–3123. [21] C.-H. Chen, P.-Y. Chiang, C.F. Jou, A low voltage mixer with improved noise figure, IEEE Microwave and Wireless Components Letters 19 (2) (2009) 92–94. [22] S.G. Lee, J.K. Choi, Current-reuse bleeding mixer, Electronics Letters 36 (8) (2000) 696–697. [23] W. Pan, J. Wang, J. Huangfu, C. Li, L. Ran, Null point elimination using RF phase shifter in continuous-wave Doppler radar system, Electronics Letters 47 (21) (2011) 1196–1198. [24] H.-R. Chuang, H.-C. Kuo, F.-L. Lin, T.-H. Huang, C.-S. Kuo, Y.-W. Ou, 60-GHz millimeter-wave life detection system (MLDS) for noncontact human vitalsignal monitoring, IEEE Sensors Journal 12 (3) (2012) 602–609. [25] Ø. Aardal, S.E. Hamran, T. Berger, J. Hammerstad, T.S. Lande, Radar cross section of the human heartbeat and respiration in the 500 MHz to 3 GHz band, in: IEEE Radio and Wireless Symposium, 2011, pp. 422–425. [26] E. Bogation, Signal Integrity—Simplified, Prentice Hall, 2004. [27] Agilent 81110A 165/330 MHz Agilent 81104 A 80 MHz Pulse/Data Generator, Reference Guide, Agilent Technologies. [28] A.A. Hafez, M.A. Dessouky, H.F. Ragai, Design of a low-power ZigBee receiver front-end for wireless sensors, Microelectronics Journal 40 (11) (2009) 1561–1568. [29] X. Wang, A. Dinh, D. Teng, L. Chen, S.B. Ko, Y. Shi, V. Dal Bello-Haas, J. Basran, An impulse based sensor for medical sensing applications, in: Proceedings of the 30th Annual International IEEE EMBS Conference, 2008, pp. 5737–5740. [30] H. Trabelsi, G. Bouzid, F. Derbel, M. Masmoudi, An 863–870 MHz spreadspectrum FSK transceiver design for wireless sensor, Microelectronics Journal 41 (8) (2010) 465–473.
Contents lists available at ScienceDirect
Microelectronics Journal journal homepage: www.elsevier.com/locate/mejo
A novel low-power transceiver topology for noncontact vital sign detection including the power management technique Chie-In Lee a,b,n, Yan-Ting Lin a, Yu-Her Chen a, Wei-Cheng Lin a a b
Department of Electrical Engineering, National Sun Yat-Sen University, No. 70 Lienhai Rd., Kaohsiung 80424, Taiwan Institute of Communications Engineering, National Sun Yat-Sen University, No. 70 Lienhai Rd., Kaohsiung 80424, Taiwan
art ic l e i nf o
a b s t r a c t
Article history: Received 19 April 2013 Received in revised form 28 August 2013 Accepted 29 August 2013 Available online 28 October 2013
In this paper, power management technique utilized in the direct down-conversion non-quadrature transceiver is presented for the low-power application of vital sign detection. The simultaneous switching noise (SSN) and overshoot and undershoot of the transient waveform distortion resulting from a pulse signal will give rise to interference with a vital sign signal. The pulse width, rise/fall time, and period of pulse bias are analyzed to mitigate the interference in this investigation. Significant issues about direct-current (DC) offset and noise confronted by the presented technique are addressed based on mathematical analysis. In radio-frequency (RF) transceiver architecture including power amplifier (PA), low-noise amplifier (LNA), and mixer, the current-reused (CRU) topology is utilized to achieve low DC power consumption. The post-layout simulation results exhibit that power consumption of the transceiver using the optimized pulse bias is reduced to 40% of the power consumption for transceiver applying the DC bias. In addition, DC offset and null detection point can be alleviated by tunable phase shifter. Crown Copyright & 2013 Published by Elsevier Ltd. All rights reserved.
Keywords: Vital sign detection Power management technique Switching noise Power amplifier
1. Introduction The noncontact Doppler radar sensor requires ultra low-power circuit and high accuracy to detect the vital sign signals for longterm health care [1–3] and life detection [4]. To improve detection accuracy, many topologies have been presented. The quadrature architecture designed for a wireless communication system [5,6] is utilized to detect the vital sign with channel selection [1,2]. In addition, the double-sideband transmission system with wide frequency tuning range [3] is presented to eliminate the null detection point. However, these structures require several active components so that the direct-current (DC) power consumption of the whole system increases [7]. The direct down-conversion nonquadrature topology can be implemented in the low-power vital sign detection system due to less active components required. In the front-end circuit design, several circuit architectures have been developed for low-power application [8–10]. Complementary structure has been presented [8] at the expense of a degraded power gain to provide low-voltage capability. To enhance the gain in low-power design, the current-reused (CRU) configuration [9,10] is proposed to merge the currents required for two transistors into a single current path such that supply current can be reduced. Power consumption is only half of the cascade topology, n Corresponding author at: Department of Electrical Engineering, National Sun Yat-Sen University, No. 70 Lienhai Rd., Kaohsiung, Taiwan 80424. Tel.: þ 886 7 525 2000 4110; fax: þ886 7 525 4199. E-mail address: [email protected] (C.-I. Lee).
and gain can be approximately the same with the result of the cascade topology. From another point of view, the power management technique can be employed to further improve efficiency [11–13]. The buck–boost converter can provide dynamic power supply to the power amplifier (PA) so as to prolong the battery life. However, the switching voltage may produce simultaneous switching noise (SSN) [14,15] as well as overshoot and undershoot. These voltage ripples should be mitigated to acceptable level in order to avoid interference with the vital sign signals. Our previous research presented a high linearity and lowpower low-noise amplifier (LNA) design with nonlinear cancellation technique [16]. In this paper, a low-power direct downconversion non-quadrature transceiver with the power management technique that is applied for the noncontact vital sign detection is presented for the first time. SSN and transient waveform distortion are analyzed and minimized to improve detection accuracy of the presented system with pulse bias. For the transceiver design, the CRU topology is employed to further reduce DC power consumption of the whole system.
2. Power management technique and vital sign detection theory The main subject of this work is to design a low-power radiofrequency (RF) transceiver for the application of vital sign detection and to maintain detection accuracy simultaneously. Fig. 1 shows novel direct down-conversion non-quadrature transceiver
0026-2692/$ - see front matter Crown Copyright & 2013 Published by Elsevier Ltd. All rights reserved. http://dx.doi.org/10.1016/j.mejo.2013.08.021
1310
C.-I. Lee et al. / Microelectronics Journal 44 (2013) 1309–1315
Fig. 1. The novel direct down-conversion non-quadrature transceiver with the power management technique and the tunable phase shifter.
with the power management technique. This is the first time that the power management technique with pulse bias is implemented in the presented PA to further decrease power consumption and to alleviate interference of the SSN, overshoot, and undershoot. Fig. 2 describes the design procedure for the presented vital sign sensor with SSN and overshoot/undershoot considered. In the active state of pulse bias, maximum output power of the PA is 9.3 dBm as the drain voltage is operated at high voltage (VH) of 2.6 V. For the sleep state, the minimum output power is 7 dBm at low voltage (VL) of 0.6 V. When the pulse signal is applied to the circuits, variation of the pulse bias will produce signal reflection at the boundary due to transmission line discontinuities [17]. The reflection leads to instantaneous change of the current and voltage at each node, causing overshoot and undershoot and hence significant ripple in the transmitted signal as shown in Fig. 3(a). Besides, SSN can be generated in power and ground planes due to the rapid change in current flowing through parasitic inductor of the power supply [14,18]. Therefore, to investigate the influence of the pulse signal on the presented system, the microstrip lines, the circuit layout, and the bonding wires of the presented system are considered in the simulation. On the other hand, a DC offset effect can occur in direct down-conversion non-quadrature architecture when a RF signal at the same frequency with the local oscillator (LO) frequency is mixed with the LO signal. Significant issues about DC offset and noise confronted by the presented technique will be addressed based on mathematical analysis. The transmitted signal of a transmitter influenced by transient waveform distortion due to pulse bias can be expressed as follows: TðtÞ ¼ AðtÞ sin ð2πf t þ φÞ
ð1Þ
where T(t) is the transmitted signal from the transmitter, f is the carrier frequency, and φ is the residual phase that is accumulated in the system. The ripple voltage of the transmitted signal due to transient waveform distortion is given by A(t). When this transmitted signal is reflected by a human subject at a nominal distance d0, the periodic body movements x(t) due to human cardiopulmonary activity will be modulated in the phase of the received signal according to the Doppler theory [2]. In addition, voltage amplitude variation still exists in the received signal along with the vital sign signal. Therefore, the received signal R(t) can be expressed as 4π 4πd0 RðtÞ ¼ Rr ðtÞ sin 2πf t xðtÞ þφ ð2Þ λ λ where Rr(t) is the amplitude of the received signal with the timevarying ripple voltage and λ is the wavelength of the RF signal. Information about the human cardiopulmonary activity can be demodulated through the mixer with an LO signal, which is derived from the transmitted signal. Note that the phase noise from LO can
Fig. 2. The flow chart describes the design process for the power management technique based on the vital sign detection.
be neglected due to range correlation [2,19]. Therefore, a detected signal in the baseband B(t) can be expressed as BðtÞ ¼ Br ðtÞ cos ðφB
4πxðtÞ Þ λ
ð3Þ
where φB ¼ φ
4πd0 φLO λ
ð4Þ
Br(t) is the amplitude of the baseband signal with the timevarying ripple voltage and φLO is the phase of the LO signal. Based on the Taylor series expansion, the baseband signal shown in Eq. (3) can be approximated as BðtÞ ¼ Br ðtÞ sin φB U
4π U xðtÞ þ Br ðtÞ cos φB λ
ð5Þ
High-order terms of the expansion are neglected under the condition xðtÞ o o λ. The first term on the right-hand side of Eq. (5) contains the vital sign signal as well as the ripple voltage at baseband Br(t). This ripple voltage can be a significant noise source, which can overwhelm the vital sign signal, degrading the detection accuracy. The second term on the right-hand side of Eq. (5)
C.-I. Lee et al. / Microelectronics Journal 44 (2013) 1309–1315
1311
Fig. 3. (a) Schematic diagram of the presented class-E PA with a stacked DPTF and a balun. Blue lines show the overshoot and undershoot of the transient waveform distortion occurred in the circuit when the pulse bias at the drain terminal of the PA is employed. (b) Layout of the class-E CRU PA. (c) Schematic diagram of the CRU LNA with a differential inductor. (d) Layout of the CRU LNA. (e) Schematic diagram of the CRU mixer. (f) Layout of the CRU mixer. (For interpretation of the references to color in this figure legend, the reader is referred to the web version of this article.)
represents the DC offset. This DC term has to be considered especially for the direct down-conversion non-quadrature architecture. In noncontact vital sign detection, the DC offset usually comes from clutter reflection and from circuit imperfection. Because the DC offset due to circuit imperfection is alleviated by high RF–LO isolation mixer [20], clutter reflection becomes the dominant factor in the Doppler vital sign detection [20]. Therefore, the source of DC offset considered in this paper is from clutter reflection. In Eq. (5), in addition to the time-varying movement of the vital sign, undesired time-varying ripple amplitude Br(t) is also included at baseband. If ripple voltage of the transmitted signal is large, detection accuracy will be significantly decreased. As a result, appropriate pulse rise/fall time is selected to avoid transient waveform distortion and to minimize amplitude variation for vital sign detection. Furthermore, to achieve accurate detection, pulse period is designed as short as possible such that
the vital sign signals can almost occur in the active state. Nevertheless, SSN will increase when a short period or rise/fall time pulse bias is adopted [14]. Therefore, as a compromise between the accurate detection and switching noise, the period and rise/fall time of the pulse bias are optimized for vital sign detection. As the pulse period and rise/fall time are fixed, the pulse width can be shortened to reduce power consumption for the presented system. On the other hand, the transmitted power is proportional to the pulse width. Therefore, selection of the pulse width is a tradeoff between overall power consumption and detection accuracy. To maintain vital sign detection accuracy with low power consumption, the pulse width is shortened such that amplitude of the vital sign signals in the baseband channel can still be 10–20 dB larger than the noise floor required for noncontact vital sign detection [19].
1312
C.-I. Lee et al. / Microelectronics Journal 44 (2013) 1309–1315
3. The transceiver design In the transmitter architecture of the presented system, the CRU topology is utilized to reduce power consumption of the PA. The output power of CRU PA is improved by using the transformer to combine the power delivered from several power cells. Fig. 3 (a) shows the architecture of a fully integrated PA operated in class-E mode consisting of the balun, the stacked double primary transformer (DPTF), and two push-pull PAs. The transistors (M1–M4) are in common-source configuration. An inter-stage network consisting of coupling capacitances C1 and C2 is designed for better signal matching. DC current flows through stacked transistors due to the low reactance of inductances (L1, L2) as shown in Fig. 3(a). Amplified RF signals from M1 and M2 are impeded by L1 and L2, and thus RF signals are fed into the gate terminal of M3 and M4 through C1 and C2. Therefore, the overall transconductance of this power stage can be equal to the value of the cascade topology. The self-bias topology is utilized in the driver stages to prevent the degradation of CMOS devices. Table 1 summarizes the post-layout simulated PA performance using CMOS 0.18 μm technology design kit provided by Taiwan Semiconductor Manufacturing Company (TSMC), Hsinchu, Taiwan. The power consumption of CRU PA is decreased to about 62.3% of conventional cascode PA, which is reduced from 47 mW to 29.3 mW, and the output power is only slightly degraded. To further enhance the output power under the condition of low power consumption, passive balun and DPTF are adopted for input and output matching of the presented PA, respectively. Input balun transforms the input single-ended signal to the differential signal as shown in Fig. 3(a). The output ports of the driver stages are connected to input ports of the power stages through capacitance and to one of the two primary coil of DPTF in order to improve the output power. In addition, the output power from power stages is fed to the other primary coil. The presented DPTF structure consisting of two three-turn primary coils and a single three-turn secondary coil is employed to enhance the output power, and current directions of the primary coils are the same to avoid self-cancellation. The insertion loss of DPTF from the silicon substrate is mitigated by connecting output port of the DPTF with an additional metal–insulator–metal capacitance. The layout of the presented class-E CRU PA with passive components is shown in Fig. 3(b). The output power of the whole PA with passive components is higher than the values of CRU PA and conventional cascode PA as shown in Table 1. Transmitted power at fundamental frequency of 2.4 GHz from the presented PA is about 9.3 dBm, with the DC power consumption of 28.1 mW. Therefore, this CRU PA with the balun and DPTF can achieve high output power and low power consumption simultaneously for remote vital sign detection or life detection of buried victims under rubble [4]. For the receiver architecture, a complementary CRU LNA and a CRU mixer are designed to achieve low power consumption. The complementary CRU LNA consists of three stages in commonsource configuration as shown in Fig. 3(c). The first stage of the CRU LNA includes a differential inductor connected at the source terminal of M1. The phase of the gate and source voltage of the M1 transistor can be opposite by using the differential inductor so
Table 1 The output power and the power consumption for different PA topologies. Topology
Conventional cascode PA
CRU PA
CRU PA with the DPTF and balun
Maximum Output power (dBm) Power consumption(mW)
4.4 47
3.2 29.3
9.3 28.1
that the transconductance of input transistor can be increased. Therefore, gain of the whole LNA circuit increases larger than the conventional CRU topology under the condition of low power consumption. In addition, the differential inductor provides inductive compensation for an input network to reduce the noise figure. The second and third stages consisting of M2 and M3, respectively, utilize a complementary topology to further reduce the DC supply voltage. L1, C1, and C2 are employed for inter-stage matching. The layout of complementary CRU LNA is shown in Fig. 3(d). Postlayout simulated gain and noise figure at 2.4 GHz are 10.1 dB and 3.4 dB, respectively. It is known that the performance of the direct down-conversion receiver may suffer from the flicker noise [21]. Therefore, in the mixer design, a CRU topology consisting of M1 and M2 transistors shown in Fig. 3(e) is utilized. With the CRU topology, current flowing through the switch stage M2 can be reduced so that flicker noise of the mixer can be mitigated [21,22]. The layout of CRU mixer is shown in Fig. 3(f). Post-layout simulation shows 5 dB and 7 dB of gain and noise figure, respectively. In receiver architecture, tunable phase shifter shown in Fig. 1 is employed to alleviate the DC offset and null detection point [23,24]. Sensitivity is a significant parameter for the receiver. To estimate sensitivity of the presented system, receiver sensitivity is determined based on the formula given as [19] Sensitivity ðdBmÞ ¼ kTB ðdBmÞ þNF r ðdBÞ þ SNR ðdBÞ
ð6Þ
where T is the temperature in kelvin, B is the receiver output bandwidth, k is Boltzmann's constant, NFr is the noise figure in the receiving path, and SNR is the signal-to-noise ratio required for the noncontact vital sign detection, which is 10–20 dB typically [19]. Assuming that the phase noise from the LO is neglected due to the range correlation [2,19], the sensitivity of the presented sensor is calculated to be 82.5 dBm using Eq. (6) at room temperature 290 K, with a bandwidth of 1 MHz [19], noise figure of 16.5 dB in the receiving path, and SNR of 15 dB within the typical range of 10–20 dB [19]. The Advanced Technical Materials 340-441-2 horn antennas with 15 dBi gain are used to transmit a 2.4 GHz signal with maximum output power of 9.3 dBm and to receive the reflected signal from the human body. The parameters of the components mentioned above are utilized in the simulator. Maximum detecting distance between the antenna and the human body can be 4.4 m approximately based on the radar equation [19] with a typical radar cross section of the human heartbeat at 2.4 GHz [25]. Therefore, the presented sensor can detect vital sign signals within 4.4 m.
4. Results and discussion Power management technique is applied to the presented vital sign detection system to further decrease power consumption. To determine the optimal pulse bias for satisfying the requirement of SSN and waveform distortion with low power consumption, output characteristics of the sensor with different pulse periods, rise/ fall time Tr,f, and width are simulated. Acceptable SSN level is required to be lower than 10% of the output voltage amplitude [26]. As for the waveform distortion, overshoot and undershoot currents have to be below 5% of the output current amplitude [27]. In Fig. 4(a), voltage fluctuation of the pulse biased sensor due to SSN decreases with increase in pulse rise/fall time. It is also observed that voltage fluctuation is almost independent of the pulse period as the pulse period is longer than 1 ms as shown in the inset of Fig. 4(a). Therefore, optimal pulse period is chosen as 1 ms as a compromise between SSN interference and accurate detection. As for the rise/fall time, the value of 60 ns is selected to satisfy the acceptable level for both SSN and overshoot/undershoot
C.-I. Lee et al. / Microelectronics Journal 44 (2013) 1309–1315
1313
Fig. 4. (a) The SSN versus rise/fall time at different pulse periods of 0.1 ms (dash–dot), 0.4 ms (dot), 0.7 ms (dash), and 1.0 ms (solid). In the inset, the SSN with the pulse periods of 1.0 ms (solid), 10 ms (dash–dot–dot), 20 ms (short dash), and 30 ms (short dot) are shown. (b) The overshoot current (circles), undershoot (squares), and output ripple voltage (triangles) versus rise/fall time at the pulse bias with VH/VL ¼ 2.6/0.6 V and 1 ms pulse period.
Fig. 5. (a) The detected signal spectrum with the different ripple voltages is shown. The optimal rise/fall time of 60 ns corresponds to the ripple voltage of 0.085 V. To purely analyze the influence of rise/fall time on the signal spectrum, the pulse width approximates to one period. (b) The vital sign signals and the DC offset versus the LO phase under the DC bias condition.
currents as shown in Fig. 4(a and b). In Fig. 5(a), we further investigate the effect of the ripple voltage resulting from overshoot/undershoot on vital sign signal spectrum at baseband due to the pulse bias applied. Noise floor increases with the increase in ripple voltage because amplitude variation due to the ripple significantly affects the baseband signals based on Eq. (5). The large amplitude of the respiration signal is not influenced by the ripple voltage. However, the significant noise floor due to large ripple voltage overwhelms the heartbeat signal as shown in Fig. 5 (a). If the ripple voltage is reduced to 0.085 V corresponding to optimal rise/fall time of 60 ns for the overshoot/undershoot requirement shown in Fig. 4(b), the accurate heartbeat signal can be obtained as shown in Fig. 5(a). This indicates that the requirement of overshoot/undershoot current can prevent the baseband signal from being influenced by noise resulting from the timevarying ripple voltage, and thus ensure reliable vital sign detection. Therefore, the optimal pulse period and rise/fall time applied to the presented CRU PA are designed to be 1 ms and 60 ns, respectively. To prevent the vital sign signal from being influenced by the DC offset and null detection point, LO phase is tuned through the tunable phase shifter. The DC offset can be reduced to 0.32 mV when the LO phase is 701, and the maximum amplitude of the vital sign signals can be achieved without the null detection point influenced as shown in Fig. 5(b). In Fig. 6(a–c), maximum vital sign signals and the minimum DC offset can be observed in different pulse width as LO phase is 701, which is consistent with the value obtained from the result with DC bias as shown in Fig. 5(b). Therefore, this demonstrates that the
phase shifter can still provide the function of LO phase tuning to increase detection accuracy even when the pulse bias is applied in the presented system. Fig. 6(b–d) exhibits a tradeoff between power consumption and the detected accuracy. When pulse width is larger than or equal to 0.3 ms, both the amplitudes of respiration and heartbeat signals fulfill the required 10–20 dB signal-to-noise ratio for the noncontact vital sign detection [19]. Therefore, to save DC power consumption without sacrificing the accuracy, optimal pulse width is selected to be 0.3 ms. In summary, optimal pulse bias condition is 1 ms period, 60 ns rise/fall time, and 0.3 ms width. This pulse bias can be provided by the commercial DC–DC converter module. To attain the performances of the sensor purely with pulse bias, the presented system without the usage of DC–DC converter is employed in the simulation environment. In Fig. 7, vital sign signals including the respiration and heartbeat signals are determined by using the presented system with optimal LO phase and pulse period, rise/fall time, and width. Therefore, power management technique not only reduces power consumption of the system but also maintains the capability of vital sign detection. Table 2 summarizes the performances of the low-power systems based on simulation. The power consumption of RF receiver and CRU PA in the presented system by using the optimal pulse bias is 6.6 mW and 8.4 mW, respectively. Therefore, low power consumption is achieved in the presented topology. Other characteristics are in acceptable ranges when compared with performances from other works [28–30]. Based on the above analysis, low power and high accuracy can be simultaneously achieved in the presented system using optimal pulse bias.
1314
C.-I. Lee et al. / Microelectronics Journal 44 (2013) 1309–1315
Fig. 6. Analysis of (a) the DC offset versus the LO phase and the pulse width, (b) the respiration versus the LO phase and the pulse width, (c) the heartbeat versus the LO phase and the pulse width, and (d) the power consumption of the PA versus the pulse width.
Table 2 The performance summary of the low power system for the sensor applications. Specification
This issue
[28]
[29]
[30]
Technology Mode
0.18 μm CMOS TX
RX
0.13 μm CMOS RX
0.18 μm CMOS TX RX
0.35 μm CMOS TX RX
TX power (dBm) Bandwidth (MHz) Noise figure (dB) Power consumption (mW)
9.3 200 4.1a 8.4/21.4c
6.6
N/A 15 8.5a 7.2
N/A 7500 3b 40
0 7 10 30
40
a
The noise figure of the cascaded system (LNA þmixer). The noise figure of the LNA. The power consumption of the proposed transceiver includes 13 mW of the AIC2354 DC/DC converter for practical purposes. b c
Fig. 7. At the LO phase of 701, the detected signal spectrum by using the optimal pulse bias. (VH/VL ¼2.6/0.6 V, Tr,f ¼60 ns, the pulse period ¼1 ms, and the pulse width ¼ 0.3 ms)
5. Conclusion In this paper, power management technique is applied to 2.4 GHz direct down-conversion non-quadrature transceiver to reduce power consumption for noncontact vital sign detection for the first time. The DC offset and noise correlated to the presented system have been
addressed by in-depth analysis. Optimal pulse period, width, and rise/fall time are determined to be 1 ms, 0.3 ms, and 60 ns, respectively. By using this optimal pulse bias, SSN can meet the requirement and the overshoot and undershoot of current can be reduced to 5% of the output current amplitude so that vital sign detection does not interfere with output voltage ripples. The CRU topology is utilized in the transceiver to achieve low power consumption. The total power consumption of the presented system using the optimal pulse
C.-I. Lee et al. / Microelectronics Journal 44 (2013) 1309–1315
bias is approximately 60% less than the result of the system using the DC bias. Furthermore, the tunable phase shifter is used to alleviate the null detection point and DC offset. Therefore, the accurate vital sign detection under low-power operating condition can be achieved by using the presented power management technique. Acknowledgments The authors would like to thank National Chip Implementation Center (CIC), Hsinchu, Taiwan for TSMC design kits, and the Wireless Communication Antenna Research Center, Kaohsiung, Taiwan for the support. This work is also supported in part by the National Science Council of Taiwan under Grant NSC101-2221E-110-077 and under Grant NSC100-2221-E-110-085. References [1] B.K. Park, O. Boric-Lubecke, V.M. Lubecke, Arctangent demodulation with DC offset compensation in quadrature doppler radar receiver systems, IEEE Transactions on Microwave Theory and Techniques 55 (5) (2007) 1073–1079. [2] A.D. Droitcour, O. Boric-Lubecke, V.M. Lubecke, J. Lin, G.T.A. Kovacs, Range correlation and I/Q performance benefits in single-chip silicon Doppler radars for noncontact cardiopulmonary monitoring, IEEE Transactions on Microwave Theory and Techniques 52 (3) (2004) 838–848. [3] Y. Xiao, J. Lin, O. Boric-Lubecke, V.M. Lubecke, Frequency–tuning technique for remote detection of heartbeat and respiration using low-power double-sideband transmission in the Ka-band, IEEE Transactions on Microwave Theory and Techniques 54 (5) (2006) 2023–2032. [4] J.B.H. Tahar, J.C. Bolomey, M. Cauterman, Microwave life detector for buried victims, in: Proceedings of the 23rd European Microwave Conference, 1993, pp. 263–265. [5] Q. Wan, C. Wang, A low-voltage low-power CMOS transmitter front-end using current mode approach for 2.4 GHz wireless communications, Microelectronics Journal 42 (5) (2011) 766–771. [6] M.A. Abdelghany, R.K. Pokharel, H. Kanaya, K. Yoshida, A low flicker noise direct conversion receiver for IEEE 802.11 g wireless LAN using differential active inductor, Microelectronics Journal 42 (2) (2011) 283–290. [7] C. Li, Y. Xiao, J. Lin, Design guidelines for radio frequency non-contact vital sign detection, in: Proceedings of the 29th Annual International IEEE EMBS Conference, 2007, pp. 1651–1654. [8] F. Roewer, U. Kleine, A novel class of complementary folded-cascode opamps for low voltage, IEEE Journal of Solid-State Circuits 37 (8) (2002) 1080–1083. [9] L. Zheng, H.C. Yao, F. Tzeng, P. Heydari, Design and analysis of a current-reuse transmitter for ultra-low power applications, in: Proceedings of the International Symposium on Circuits Systems, 2009, pp. 1317–1320. [10] H.-H. Hsieh, L.-H. Lu, Design of ultra-low-voltage RF frontends with complementary current-reused architectures, IEEE Transactions on Microwave Theory and Techniques 55 (7) (2007) 1445–1458. [11] B. Sahu, G.A. Rincon-Mora, A high-efficiency linear RF power amplifier with a power-tracking dynamically adaptive buck-boost supply, IEEE Transactions on Microwave Theory and Techniques 52 (1) (2004) 112–120.
1315
[12] F.H. Raab, B.E. Sigmon, R.G. Myers, R.M. Jackson, L-band transmitter using Kahn EER technique, IEEE Transactions on Microwave Theory and Techniques 46 (12) (1998) 2220–2225. [13] M. Iwamoto, A. Williams, P.-F. Chen, A.G. Metzger, L.E. Larson, P.M. Asbeck, An extended Doherty amplifier with high efficiency over a wide power range, IEEE Transactions on Microwave Theory and Techniques 49 (12) (2001) 2472–2479. [14] M.A. Méndez, J.L. González, D. Mateo, A. Rubio, An investigation on the relation between digital circuitry characteristics and power supply noise spectrum in mixed-signal CMOS integrated circuits, Microelectronics Journal 36 (1) (2005) 77–84. [15] K.B. Unchwaniwala, M.F. Caggiano, R.C. Frye, Noise generation, propagation and effects on RF components in system-on-a-chip packaging, Microelectronics Journal 33 (5-6) (2002) 471–478. [16] C.I. Lee, W.C. Lin, J.M. Lin, Low-power and high-linearity SiGe HBT low-noise amplifier using IM3 cancellation technique, Microelectronics Engineering 91 (2012) 59–63. [17] R. Achar, M.S. Nakhla, Simulation of high-speed interconnects, Proceedings of the IEEE 89 (5) (2001) 693–728. [18] A. Kabbani, A.J. Al-Khalili, Estimation of ground bounce effects on CMOS circuits, IEEE Transactions on Components, Packaging and Manufacturing Technology 22 (2) (1999) 316–325. [19] C. Li, X. Yu, C.-M. Lee, D. Li, L. Ran, J. Lin, High-sensitivity software-configurable 5.8-GHz radar sensor receiver chip in 0.13-μm CMOS for noncontact vital sign detection, IEEE Transactions on Microwave Theory and Techniques 58 (5) (2010) 1410–1419. [20] C. Gu, R. Li, H. Zhang, A.Y.C. Fung, C. Torres, S.B. Jiang, C. Li, Accurate respiration measurement using DC-coupled continuous-wave radar sensor for motionadaptive cancer radiotherapy, IEEE Transactions on Biomedical Engineering 59 (11) (2012) 3117–3123. [21] C.-H. Chen, P.-Y. Chiang, C.F. Jou, A low voltage mixer with improved noise figure, IEEE Microwave and Wireless Components Letters 19 (2) (2009) 92–94. [22] S.G. Lee, J.K. Choi, Current-reuse bleeding mixer, Electronics Letters 36 (8) (2000) 696–697. [23] W. Pan, J. Wang, J. Huangfu, C. Li, L. Ran, Null point elimination using RF phase shifter in continuous-wave Doppler radar system, Electronics Letters 47 (21) (2011) 1196–1198. [24] H.-R. Chuang, H.-C. Kuo, F.-L. Lin, T.-H. Huang, C.-S. Kuo, Y.-W. Ou, 60-GHz millimeter-wave life detection system (MLDS) for noncontact human vitalsignal monitoring, IEEE Sensors Journal 12 (3) (2012) 602–609. [25] Ø. Aardal, S.E. Hamran, T. Berger, J. Hammerstad, T.S. Lande, Radar cross section of the human heartbeat and respiration in the 500 MHz to 3 GHz band, in: IEEE Radio and Wireless Symposium, 2011, pp. 422–425. [26] E. Bogation, Signal Integrity—Simplified, Prentice Hall, 2004. [27] Agilent 81110A 165/330 MHz Agilent 81104 A 80 MHz Pulse/Data Generator, Reference Guide, Agilent Technologies. [28] A.A. Hafez, M.A. Dessouky, H.F. Ragai, Design of a low-power ZigBee receiver front-end for wireless sensors, Microelectronics Journal 40 (11) (2009) 1561–1568. [29] X. Wang, A. Dinh, D. Teng, L. Chen, S.B. Ko, Y. Shi, V. Dal Bello-Haas, J. Basran, An impulse based sensor for medical sensing applications, in: Proceedings of the 30th Annual International IEEE EMBS Conference, 2008, pp. 5737–5740. [30] H. Trabelsi, G. Bouzid, F. Derbel, M. Masmoudi, An 863–870 MHz spreadspectrum FSK transceiver design for wireless sensor, Microelectronics Journal 41 (8) (2010) 465–473.