1822
IEEE TRANSACTIONS ON BIOMEDICAL ENGINEERING, VOL. 54, NO. 10, OCTOBER 2007
A Biosignal Instrumentation System Using Capacitive Coupling for Power and Signal Isolation Kari Väinö Tapio Piipponen*, Member, IEEE, Raimo Sepponen, Member, IEEE, and Pekka Eskelinen, Member, IEEE
Abstract—Requirements for patient safety and a high interference rejection ratio in medical equipment create a demand for effective isolation devices. A system scale approach that uses capacitive coupling for power and signal isolation is presented. In addition, we describe the development of an instrumentation system prototype that applies microwaves for power exchange and bidirectional data transfer across the isolation barrier. The system consists of an isolated transducer unit, a central unit, and a single coaxial cable between the units. The isolation capacitance is as low as 1.6 pF, inclusive of the digital data transfer and power exchange up to 600 mW of isolated direct current (dc) power. The system is suitable for line-powered biopotential measurements and it is shown that reducing the isolation capacitance from 180 to 1.6 pF improves the power line rejection by 30 dB in a typical electrocardiogram (ECG) measurement setup. Index Terms—Biopotential, common mode interference, power isolation, power line interference, signal isolation.
I. INTRODUCTION
B
IOPOTENTIAL signals such as electrocardiogram (ECG), electromyography (EMG), and electroencephalography (EEG) are often recorded in the presence of electromagnetic interference. Common mode interference and its coupling mechanisms from power lines to biopotential measurement systems are presented elsewhere [1]–[3]. The input-referred is interfering voltage due to common mode voltage mainly determined by the potential divider effect caused by the essential differences in impedance between individual elecin respect to trodes. This is the main reason to reduce the and it is done in two ways ground-referred body voltage (Fig. 1) as follows. 1) A neutral electrode is connected between the body and the amplifier to provide a low-impedance path between the patient and the isolated ground. A driven-right-leg (DRL) circuit is typically applied to further lower the effective impedance of the neutral electrode [4], [5]. 2) Isolation devices are used to provide data and power transfer and the galvanic insulation between the amplifier and the grounded electronics. This is mainly for safety reasons [6], but the isolation capacialso determines the induced . tance Manuscript received April 24, 2006; revised January 24, 2007. Asterisk indicates corresponding author. *K. V. T. Piipponen is with the Applied Electronics Laboratory, Department of Electrical and Communications Engineering, Helsinki University of Technology, P.O. Box 3000, Otakaari 5 A, 02015 Espoo, FI-02015 TKK, Finland (e-mail:
[email protected]). R. Sepponen and P. Eskelinen are with the Applied Electronics Laboratory, Department of Electrical and Communications Engineering, Helsinki University of Technology, 02015 Espoo, Finland (e-mail:
[email protected];
[email protected]). Digital Object Identifier 10.1109/TBME.2007.894830
Fig. 1. Typical biopotential measurement setup with the neutral electrode and isolation device.
The commercially available isolation device lineup includes analog isolation amplifiers, isolated digital couplers, optocouplers, and isolated direct current (dc/dc) converters. Some of the analog isolation amplifiers and digital couplers also include small levels ( 50 mW) of isolated power [7], [8]. Inside the isolation device, a number of barrier arrangements and signal modulation schemes are available. The three techniques in common use for on-chip signal isolation are optical isolation, isolation, and isolations. In addition, optic fibers and radio transceivers are applied for signal transfer in battery powered systems. Power isolation is typically accomplished by a transformer coupling inside a dc/dc converter, but also an optical power supply [9] and on-chip capacitive coupling power exchange [10] are suggested [11]. Regardless of the coupling technology, there is a finite amount of capacitance between the input and output of all is the total capacitance of all the isolation devices. The parallel isolation devices and it determines the alternating current (ac) characteristics of the isolation barrier. For example, the input–output capacitance of a commercially available isolated dc/dc converter is typically 15–150 pF. Fig. 2(a) presents a conventional way to set up a patient monitoring system. Electrode leads are collected to a trunk cable, which connects to the amplifier input with the patient cable. The amplifier modules with isolation are placed in a chassis that provides a connection to the display and user interface via common data bus.
0018-9294/$25.00 © 2007 IEEE
PIIPPONEN et al.: BIOSIGNAL INSTRUMENTATION SYSTEM USING CAPACITIVE COUPLING
1823
conditioning parameters. The solution described in this paper involves using Bluetooth as the communication technology, because Bluetooth is already at an advanced level of development. The authors chose a Bluetooth module, including radio and baseband hardware and the protocol stack (WRAP THOR™ 2022-1, Bluegiga Technologies, Espoo, Finland). Also, Bluetooth serial port profile (SPP) with maximum data rate of 230 kb/s is implemented inside the module (iWRAP 2.1.0, Bluegiga Technologies). Any other suitable technology could be used as well. The system shown in Fig. 3 uses a point-to-point Bluetooth link for bidirectional communications between the units. Even though technology that is meant for wireless devices (Bluetooth) is applied to communications, we want to emphasize that the system is not wireless. The coaxial cable provides a very lowloss radio path for Bluetooth radio link and power exchange and connects the central unit and the transducer unit together. Fig. 2. Diagram of (a) a typical bedside patient monitor with passive lead trunk and (b) the new solution using the transducer unit with isolation.
In this paper, we will describe new technology that allows bidirectional data transfer and power exchange across an isolation barrier using capacitive coupling. Further, the development of instrumentation for measuring biopotentials using the new technology is described. Also, we will measure the amount of isolated dc power, evaluate the system’s suitability for biopotential measurements and its capability to reduce common mode interference in a typical ECG measurement setup. II. SYSTEM DESCRIPTION AND DEVELOPMENT A. Architecture A top-level diagram of the new instrumentation system including the patient is shown in Fig. 2(b). The system consists of two main parts: the central unit and the remote biomedical transducer unit with isolation. The patient is connected to the isolated transducer unit with electrode leads. In the transducer unit, the measured biopotential signal is conditioned, sampled, and transmitted to the central unit through a coaxial cable using a radio transceiver. The transducer unit has also a control circuit for adjusting the signal conditioning parameters. In the central unit, the radio signal is received with another radio transceiver and further transmitted to the display and user interface device, such as a pesonal computer (PC). The operating power for the transducer unit is generated in the central unit and transmitted to the transducer unit through the same coaxial cable using microwaves. Because microwaves are used for both data and power transfer, capacitive isolation barrier with very low amount of coupling capacitance can be used without significant attenuation of the signal. A block diagram of the system is presented in Fig. 3. B. Data Transfer Bidirectional data transfer across the isolation barrier is arranged for biosignal transfer and simultaneous control of signal
C. Power Exchange In addition to data transfer, operating power for isolated electronics must be provided. Microwaves are used also for power transfer to minimize the insertion loss of the barrier coupling capacitors, and consequently, to further improve the efficiency of the power feed. A continuous wave personal communication services (PCS) band (1.98 GHz) carrier is generated by a voltage controlled oscillator (VCO) (ROS-2500, Mini-Circuits, Brooklyn, NY). The VCO is followed by a custom power amplifier (PA) in the central unit (Fig. 3) resulting in a medium power (33 dBm) carrier at the central unit output. The power amplifier consists of a voltage variable attenuator (VACC-22, Mini-Circuits), a driver amplifier (VNA-25, Mini-Circuits), and a power amplifier (RF3805, RF Micro Devices, Greensboro, NC). The carrier is fed to the transducer unit through the same coaxial cable and the coupling capacitors that are used in data transfer. Any other frequency bands could be used as well. The authors chose the PCS band because of good availability and low cost of the commercially available components. D. Barrier Arrangement The isolation barrier is arranged by cutting both the signal and the return conductors by commercially available, lumped 0.8-pF capacitors (C17 series, Dielectric Laboratories, Cazenovia, NY) with voltage rating of 1000 V. The coupling capaciis tors are placed in parallel in the transducer unit and the the sum of these capacitors (1.6 pF). Since the isolation barrier forms the physical interface between the 50- coaxial cable and the associated 50- microstrip circuitry in the transducer unit it should be matched to 50 as well. The input impedance of the coupling capacitors is compensated with ceramic radio-frequency (RF) chip inductors (0603CS series, Coilcraft, Cary, IL), that were placed in series with the capacitors (Fig. 4). This results as a series resonance circuit that was designed and simulated with a circuit simulator (APLAC 8.00, APLAC Solutions Corporation, Espoo, Finland). The series resonance frequency was tuned as close to the power feeding frequency (1.98 GHz) as possible to minimize the barrier attenuation.
1824
IEEE TRANSACTIONS ON BIOMEDICAL ENGINEERING, VOL. 54, NO. 10, OCTOBER 2007
Fig. 3. Block diagram of the new instrumentation system.
Fig. 4. Crcuit diagram of the transducer RF circuitry including the isolation barrier, the diplexer, and the RF/dc converter.
E. Diplexers Because a single coaxial cable is used for data transfer and power feed, the power signal (1.98-GHz PCS band) and the Bluetooth signal [2.45-GHz industrial, scientific, and medical (ISM) band] can be separated by difference in frequency with specially designed diplexers (Fig. 4). To achieve simple, but effective implementation, the generic diplexer circuit was reduced, and a basic t-junction with parallel resonance circuit at the branching line (Fig. 4) was designed and simulated with a circuit simulator (APLAC). The parallel resonance frequency was also tuned as close to the power feeding frequency (1.98 GHz) as possible. F. RF/dc Converter The microwave power is converted to dc power with a specially designed RF/dc converter. The RF/dc converter consists of an input-matching circuit, a zero-bias diode rectifier in a voltage doubler configuration, and an output bypass filter (Fig. 4). Commercially available HSMS-2702 Schottky diode pair was chosen as the rectifier because of its low series resistance (0.65 ) and low junction capacitance (6.7 pF) [12]. The input-matching circuit and the output bypass filter were implemented with commercially available lumped capacitors (C06 series, Dielectric Laboratories) and RF chip inductors (0603CS series, Coilcraft). The RF/dc converter was designed and simulated with circuit simulator (APLAC) using nonlinear harmonic balance (HB)
Fig. 5. Return loss and the output dc power of the RF/dc converter partial prototype as functions of dc load.
analysis method and the design was verified by RF/dc converter partial prototype. The scalar reflection coefficient and the output ) with dc power were measured as a function of dc load 1-W incident 1.98-GHz microwave power with a custom measurement setup (Fig. 8). The simulated and measured results are shown in Fig. 5.
PIIPPONEN et al.: BIOSIGNAL INSTRUMENTATION SYSTEM USING CAPACITIVE COUPLING
1825
B. Measurement System Evaluation Setup
Fig. 6. PCB layout of the RF part of the transducer unit. For explanation of the numbers, see Section III-A.
III. EXPERIMENTAL A. System Prototype Prototypes of the transducer unit and the central unit were built according to the block diagram in Fig. 3. The central unit was implemented as a single printed circuit board (PCB) containing the PCS-band power generator, the diplexer, the Bluetooth module, power management, and universal serial bus (USB) circuitry. To avoid thermal problems, a metallic enclosure (260 170 65 mm ) was chosen to ensure sufficient amount of heat transfer capacity for the RF power amplifier, which is the most power consuming part in the central unit. The central unit was powered by a grounded, commercially available wall adapter and connected to a laptop PC by a USB cable for communications with the user interface. The transducer unit prototype consists of two PCBs. The core technology of the transducer unit, including the isolation barrier, the diplexer, the RF/dc converter, and the Bluetooth module was implemented on a single PCB with a microcontroller ( C) and power management circuitry. The PCB layout of the RF part of the transducer unit is shown in Fig. 6. The coaxial cable connects to the PCB with commercially available RF connector (1). The coupling capacitors and the serial inductors (2) transfer the RF signals (power and data) over the isolation barrier. The diplexer (3) separates the signals by difference in frequency guiding the data signal to the Bluetooth module (4) and the power signal to the RF/DC converter (5). A 16-b analog-to-digital (A/D) converter, the control circuitry, and the biopotential amplifier were implemented on another PCB, which is connected to the main transducer PCB by board-to-board connectors. The biopotential amplifier, in this case, is a custom ECG amplifier (Fig. 7). The resulting ECGare sampled by the analog-to-digital converter signal and the (ADC). The whole transducer unit prototype is encapsulated in a plastic enclosure (135 65 30 mm ). A commercially available coaxial cable (RG-58) with length of 3 m (typical length for a commercially available patient cable) was chosen to connect the transducer unit and the central unit together and to provide the path for data transfer and power feed. A custom user interface application was developed with Labview 8.1. graphical development environment (National Instruments, Austin, TX). The measured data (ECG and common mode signals) are displayed on the screen and stored to a file. In addition, the user interface provides controls for remote signal conditioning parameter adjustment. Amplifier gain, offset level, and sampling speed can be varied during the measurement.
The power feed performance of the system was evaluated and the RF/dc power supply was actually used to power up the transducer unit during the evaluation period. The amount of transferred dc power, internal noise of the system, the real ECG signal, and the induced common mode voltage were measured. The propagating and reflected RF power levels were measured with a 20-dB directional coupler (ZABDC20-2400, Mini, JFW Industries, Circuits) and RF power detectors ( Indianapolis, IN) connected between the central unit output and the coaxial cable (Fig. 8). The power detectors were calibrated with a power meter (437B, Agilent Technologies, Santa Clara, CA) and a power sensor (8481A, Agilent) to achieve absolute power levels. In the transducer unit, the RF/dc converter output can be connected to transducer unit power management circuitry to actually power up the transducer unit or optionally to resistive load (Fig. 4) to measure the absolute amount of transferred dc power with a regular voltmeter. The measurement setup proposed in [5] was applied for common mode test. A ground-referred, capacitively coupled common mode source is connected to electrode leads with was chosen resistive electrode impedances [Fig. 9(a)]. much larger (200 nF) than its typical value (2 pF) to ensure , especially at low frequencies. sufficient signal level of was chosen 100 k and was chosen 200 pF to represent the typical situation [4]. IV. RESULTS AND DISCUSSION A. Isolated dc Power First, the return loss of the transducer unit and the coaxial cable [device under test (DUT) in Fig. 8] was measured as a function of frequency (Fig. 11). Second, the amount of isolated dc power available was measured. The output of the RF/dc converter in the transducer unit was connected to resistive load (56 ) and the output dc power was measured as a function of output RF power of the central unit at 1.98-GHz frequency. As the power curve in Fig. 12 shows, up to 900 mW of dc power can be momentarily transferred across the 1.6-pF isolation barrier to 56 load in the transducer unit. The conversion efficiency of the RF/dc converter is [13] (1) From Fig. 5, we get return loss of 9.5 dB at 56- load, which correspond to 111-mW reflected RF power when the incident RF power is 1 W. The corresponding output dc power is 630 mW. Now, substituting these values to (1), we get 71% conversion efficiency for the rectifier at 56- load and 30-dBm, 1.98-GHz incident RF power. In long term use, the amount of dc power is limited by the specification of diode rectifier absolute maximum power dissipation (250 mW, [12]). Comparing that to conversion efficiency, we get the maximum output power of 612 mW for the RF/dc converter. This corresponds to little more than 32 dBm of central unit output RF power (Fig. 12).
1826
IEEE TRANSACTIONS ON BIOMEDICAL ENGINEERING, VOL. 54, NO. 10, OCTOBER 2007
Fig. 7. Single supply ECG amplifier used in system evaluation. POT-A and POT-B are digital potentiometers, so the signal offset level and amplifier gain can be adjusted in real time with the measurement.
Fig. 8. Test setup for the power feed evaluation.
B. Internal Interference The system noise was evaluated by connecting 0.5-m electrode cables to the amplifier input and connecting the leads together with 100-k resistors. A random noise voltage with maximum peak-to-peak amplitude up to 1.25 mV was seen in the sampled ECG amplifier output signal. The total gain of the ECG amplifier was set to 600, which leads to input-referred noise voltage. The internal noise 2 is likely caused by reduced (13 b) effective resolution of the ADC at 2.56-V span, the amplifier noise, and the nearby RF/dc converter.
Fig. 9. (a) Common mode coupling circuit. (b) Electrode placement for the real ECG measurement.
C. Induced Common Mode Voltage The induced can be calculated using a circuit model to analyze capacitive coupling to the patient [14] in which case the induced common mode voltage becomes
(2)
PIIPPONEN et al.: BIOSIGNAL INSTRUMENTATION SYSTEM USING CAPACITIVE COUPLING
1827
Fig. 13. ECG signal recorded with the new instrumentation system between the electrode locations V2 and V4. Fig. 10. Matching of the transducer unit (and the cable) as a function of frequency around the PCS band.
Fig. 11. Converted dc power in the transducer unit as a function of the central unit RF generator output power.
the amount of induced in respect to ground-referred body voltage . In theory, using 1.6-pF isolation capacitance instead of 180 pF improves the common mode rejection by about 40 dB over the whole frequency band. The experimental result was about 30 dB improved (Fig. 12), which is nearly equal to the effect of a classic DRL-circuit [3], [4]. We assume that there is a finite amount of stray capacitance between the isolated transducer unit and the earth ground in the , test setup. This stray capacitance would be in parallel to degrading the common mode performance by about 10 dB in . comparison to theoretical calculations in the case of small This is an interesting result, because the same effect is likely to be generally present in all instrumentation, but rarely discussed in the literature. D. Recorded ECG Signal Fig. 13 shows the ECG signal measured between the electrode locations V2 and V4 [Fig. 9(b)] with the prototype system. The neutral electrode was connected above the right hip. The measurement was performed with unshielded 0.5-m electrode cables. The offset level and the gain of the amplifier were adjusted to get a proper signal level. The bandwidth of the measurement was 0.05–150 Hz with total gain of 600. The signal was sampled with 16-b resolution, 2.56-V ADC span, and a 1-kHz sampling rate. The power line interference is not seen in the signal, but the random noise is observed during the diastole phase of the cardiac cycle. V. CONCLUSION
Fig. 12. Effect of total isolation capacitance on induced V
.
when is assumed and is the ground-referred voltage induced to the patient. was measured by connecting the common The induced mode generator [Fig. 9(a)] to the ECG cables of the transducer unit. Additional capacitors were placed in parallel to the acand the measurement tual coupling capacitors to vary the was performed as a function of frequency. As the results in Fig. 13 show, the amount isolation capacitance affects greatly
The concept of the new solution that provides an interface for isolated data and power transfer using capacitive coupling is proven with a fully functional prototype. Isolation capacitance of 1.6 pF inclusive of the bidirectional data transfer and power exchange was achieved. Advantages of the new instrumentation solution are improved patient safety and common mode rejection due to very low barrier capacitance and still reasonable amount (600 mW) of isolated dc power. The specifications of the system are summarized in Table I. We suggest that the system has plenty of potential targets for applications in the field of bioinstrumentation. Further development will focus on increasing the voltage rating of the barrier to achieve medical grade isolation and optimization of the
1828
IEEE TRANSACTIONS ON BIOMEDICAL ENGINEERING, VOL. 54, NO. 10, OCTOBER 2007
TABLE I SYSTEM SPECIFICATION SUMMARY
microwave power supply to reduce the internal electromagnetic interference (EMI) and to increase the amount of available isolated dc power and power transfer efficiency. Also, the size of the isolated transducer unit will be reduced to enable seamless integration to the patient cable. REFERENCES [1] J. C. Huhta and J. G. Webster, “60-Hz Interference in electrocardiography,” IEEE Trans. Biomed. Eng., vol. BME-20, no. 2, pp. 91–101, Mar. 1973. [2] B. B. Winter and J. G. Webster, “Reduction of interference due to common mode voltage in biopotential amplifiers,” IEEE Trans. Biomed. Eng., vol. BME-30, no. 1, pp. 58–62, Jan. 1983. [3] A. C. Metting van Rijn, A. A. Peper, and C. A. Grimbergen, “High quality recording of bioelectric events. Part I interference reduction theory and practice,” Med. Biol. Eng. Comput., vol. 28, pp. 389–397, 1990. [4] B. B. Winter and J. G. Webster, “Driven-right-leg circuit design,” IEEE Trans. Biomed. Eng., vol. BME-30, no. 1, pp. 62–66, Jan. 1983. [5] E. M. Spinelli, N. H. Martinez, and M. A. Mayosky, “A transconductance driven-right-leg circuit,” IEEE Trans. Biomed. Eng., vol. 46, no. 12, pp. 1466–1470, Dec. 1999. [6] C. H. Small, “Medical devices demand stringent isolation techniques,” Electron. Design, Strategy, News, pp. 41–49, Sep. 28, 2006.
[7] Analog Devices, Inc., AD215 Datasheet Nashau, NH, 1996. [8] B. Chen, “iCoupler products with isopower technology: Signal and power transfer across isolation barrier using microtransformers,” Analog Devices, Inc., 2006. [9] M. F. Hajer, A. C. Metting VanRijn, and C. A. Grimbergen, “Design of an optical power supply for biopotential measurement systems,” in Proc. 1st Joint BMES/EMBS Conf., 1999, vol. 2, pp. 845–. [10] E. Culurciello and A. G. Andreou, “Capacitive coupling of data and power for 3D silicon-on-insulator VLSI,” in Proc. IEEE Int. Symp. Circuits Syst., 2005, vol. 4, pp. 4142–4145, 2005. [11] P. Pickering, “A system designer’s guide to isolation devices,” in Sens. Mag., Jan. 1999. [12] Agilent Technologies, HSMS-2702 Datasheet Santa Clara, CA, 2005. [13] J. P. Curty, N. Joehl, F. Krummenavher, C. Dehollain, and M. J. Declerq, “A model for -power rectifier analysis and design,” IEEE Trans. Circuits Syst. I, Reg. Papers, vol. 52, no. 12, pp. 2771–2779, Dec. 2005. [14] M. Fernandez and R. Pallas-Areny, “A comprehensive model for power line interference in biopotential measurements,” IEEE Trans. Instrum. Meas., vol. 49, no. 3, pp. 535–540, Jun. 2000.
Kari Väinö Tapio Piipponen (S’06–M’07) was born in Kerava, Finland, in 1977. He received the M.S. degree in electrical engineering from Helsinki University of Technology, Helsinki, Finland, in 2003. In 1997, he was promoted to the rank of second lieutenant in the Finnish army reserve. In 1998, he joined the Department of Electrical and Communications Engineering, Helsinki University of Technology, as a student, where he is currently working as a Research Scientist at the Applied Electronics Laboratory. His main interests are biomedical instrumentation, embedded systems, and short-range radio communications. Mr. Piipponen is a member in the Finnish Society of Electronics Engineers, the Finnish Association of Graduate Engineers, and the Association of Electrical Engineers in Finland.
Raimo Sepponen (S’72–M’81), photograph and biography not available at the time of publication.
Pekka Eskelinen (M’94), photograph and biography not available at the time of publication
