Integration of Supercapacitors into Wirelessly Charged Biomedical Sensors

Transcription

1 Integration of s into Wirelessly Charged Biomedical Sensors Amit Pandey, Fadi Allos, Aiguo Patrick Hu, David Budgett The Department of Electrical and Computer Engineering The University of Auckland Auckland, New Zealand apan038@aucklanduni.ac.nz Abstract Conventional implantable sensors are often battery operated. Batteries are characterized by their high energy densities which allow for long term operation. However, it can take a long time to charge up a battery. As an emerging technology, a supercapacitor can be charged very quickly and it is considered as alternative energy storage to replace traditional batteries. This paper proposes an implantable biomedical sensor which operates off a super capacitor which is wirelessly charged. A feasibility analysis of supercapacitors is presented first, and then the system design and implementation are detailed for achieving wireless charging control with digital data logging. Design considerations relating to supercapacitor specific issues are also presented in detail. The implemented circuit can wirelessly charge a 5.5V, 1F supercapacitor in approximately 5 seconds. This capacitor powers a microcontroller and wireless transceiver for approximately 60 minutes before the circuit needs to be charged again. Keywords-Biomedical sensors; wireless power; supercapcitors;inductive power transfer. I. INTRODUCTION The continued miniaturization of electronic circuitry is an enabling technology for implantable biomedical sensors. Implantable sensors typically operate from batteries. Despite having high energy densities [1], charging times of several hours can be inconvenient when used in implantable devices. Recent advancements in supercapacitor technology have blurred the functional distinction between capacitors and batteries, with capacitors having increasingly large capacities. Although the energy density of supercapacitors is still comparatively low [1], the possibility of using supercapacitors as a replacement for batteries now exists. This prospect is most attractive in applications where fast charging and minimal charge circuitry are necessary. Traditional applications for supercapacitors have included using them as a source for peak output power for battery operated devices [2]. In addition to this, supercapacitors are commonly used as energy harvesting buffers. Here, the supercapacitor is charged from an energy harvester such as a solar cell [3], and supplements power delivery into a load. The intention of this research is to investigate the potential for a supercapacitor to be used as a battery replacement in low power biomedical sensor applications. This will be supplemented with the development of a wireless charging system for the supercapacitor, which utilizes Inductive Power Transfer (IPT) technology to wirelessly charge the supercapacitor. The paper is organized as follows: the relevant background of supercapacitors and biomedical sensor power requirements is initially presented. This includes a discussion of the traditional approaches to powering implantable sensors, and the advantages supercapacitors have over these existing approaches. A simple model for estimating charge and discharge times of supercapacitors is also presented. A wireless charging system based on IPT is then proposed. A discussion of elements of IPT and how they are applied to supercapacitor charging, as well as a wireless charge control algorithm, a hybrid DC-DC converter topology, and other additional aspects of the final implementation are presented and discussed in detail. II. BACKGROUND Traditionally, implantable sensors have been battery operated. Batteries enable the circuit to be encapsulated, sealed and selfcontained and, with careful power management, can generally sustain circuit operation for relatively long periods of time. However, there are a number of disadvantages associated with batteries: Secondary batteries have long charge times. An implantable sensor produced by Telemetry Research Ltd requires 4 hours to charge, and will provide 9 hours of discharge time [4]. Certain varieties of batteries are prone to memory effects. This means that the battery can be degraded over time if not adequately maintained. Batteries contain many toxic heavy metals. s are presently used in a wide range of secondary power source applications. These applications generally involve the supercapacitor supplementing a conventional battery source during peak load conditions [5]. The justification for using supercapacitors in such a manner comes down to the distinction between energy density (energy stored per unit volume) and power density (charge and discharge times).

2 employed. The converter would ensure a constant output voltage, regardless of the input supercapacitor voltage. For the purposes of the model, it was also assumed the DC-DC converter is 100% efficient. These assumptions allowed for the derivation of this formula for discharge time: (2) Figure 1: Comparison of conventional energy sources [1] s can be charged and discharged over very short periods of time (i.e. they have high power densities), thus they are suited to applications which require sporadic pulses of current. However, supercapacitors have a relatively low energy density which means batteries are currently preferred in applications requiring constant power over longer periods of time [1]. In addition to having fast charging characteristics, supercapacitors have further advantages over batteries: Can easily determine a capacitors state of charge by measuring its terminal voltage. Capacitors can tolerate extremely high charge and discharge currents. Capacitors are generally RoHS (Restriction of Hazardous Substances Directive) compliant with low hazard risk. A. Modeling supercapacitor characteristics A model was developed to estimate the charge and discharge times of various supercapacitor varieties. The model is shown in Figure 2 below. DC/DC Converter Figure 2: charge and discharge model It was assumed that a constant DC current would be used to charge the supercapacitor. This assumption allowed for the derivation of the following formula: The second component of the model was developed to estimate how long it would take for a supercapacitor to discharge. It was assumed that a DC-DC converter would be (1) V MIN is the minimum input voltage into the DC-DC converter for which the output voltage can be sustained. In order to maximise the time for which the supercapacitor can maintain operation, (V RATED 2 V MIN 2 ) should be maximised. Methods to achieve this will be discussed later in the paper. For the purposes of this research, a Cap-xx branded 5.5V, 1F supercapacitor was utilized. This selection was based primarily on the small profile of these supercapacitors. Equation two above indicated that this supercapacitor could sustain circuit approximation for approximately 40 minutes with a load current of 2mA, and a load voltage of 3V. III. IPT CHARGING SYSTEM Recharging is an important feature for implantable sensors required for lifetime operation; the alternative is surgery to replace a battery. IPT is a technique which allows the wireless transfer of power between two magnetically coupled systems [6] this enables power to be transferred across the skin to a fully sealed implanted device. An IPT system generally consists of a primary and secondary side. Both contain and are wirelessly coupled by inductor windings. A high frequency alternating current can be generated in the primary which produces an alternating magnetic field. This magnetic field is coupled to the secondary inductor winding resulting in an induced voltage dictated by Faraday s Law. The degree of magnetic coupling between the primary and secondary sides is generally quite low. This means that the induced voltage across the secondary inductor coil is unsuitable to supply a load directly. Thus, some form of power regulation and conditioning must be employed on the secondary to provide a suitable voltage [6]. A. Primary resonant converter Fundamental to the operation of any IPT system is the generation of a high frequency alternating current. It is this high frequency alternating current which produces a magnetic field that can couple from one system to another. The generation of this high frequency alternating current occurs on the primary side of any IPT topology. In this research, only resonant DC-AC converters have been considered. Resonant converters are classified according to whether their input is a voltage source (voltage-fed) or current source (current-fed). Voltage fed converters may be full bridge (four switching devices) or half bridge (two switching devices, and two capacitors). Similarly, a current fed converter could be full bridge (four switching devices) or push pull (where a

3 phase split inductor is used to split the DC current replacing two of the switches). A current source is mimicked by placing a large inductance in series with a DC voltage source [6]. The primary track inductance distributes the magnetic flux such that it can couple with the secondary and induce a corresponding voltage. Without adequate tuning (in the form of a parallel or series capacitance) it may be difficult to drive a large track inductance at high frequencies [7]. The implemented topology is a push-pull current fed resonant converter. If a full bridge topology was employed, isolated gate drivers would be required as the negative rail of the DC supply is to be used as the common ground. Further to this, the push-pull operation results in a resonant voltage which is double that of the full bridge topology. It can be shown [7] that: Where is the primary track current, is the DC input voltage, is the resonant frequency, is the primary inductance and is the Q factor. In general, Zero Voltage Switching (ZVS) can be achieved by either dynamically altering the switching frequency, or the tuning of the LC tank circuit. These schemes however require external controllers to be implemented. With a slight modification of the selected current fed topology, start up and sustained ZVS operation can be achieved without the aid of any external controller [6]. The implemented topology is depicted in Figure 4 below. Start up and steady state ZVS operation can be guaranteed provided the Q factor of the converter exceeds 2.54, [6] where: (3) across one MOSFET. It can be seen that as the resonant voltage decreases, so does the gate drive. Figure 3: Uncompensated pickup voltage (sinusoid) and gate voltage of designed prototype. Demonstrates ZVS. For this prototype, = 15, p = 300nF, Ls = 2.2uH B. Dynamic switching of resonant converter In order to dynamically switch the resonant converter on and off, a low side switch has been added to the resonant converter described above. The implemented topology is shown in Figure 4 below. This switch is controlled by a microcontroller on the primary side. The purpose of this microcontroller is to switch the resonant converter on and off depending on the state of the supercapacitor. This feature has been implemented primarily to reduce the amount of circuitry on the secondary side. The development of this digital control scheme is detailed later in the paper. LDC LPS LPS (4) R R Where reflected 2⁰ impedance. During steady state operation, the current is split equally down the two legs of the converter. At any instant, half the current will flow through the LC tank, whilst the other half is sunk directly to ground. It is the current flowing through the LC tank which results in a sinusoidal waveform. As described above, with the implemented topology steady state ZVS operation can be achieved without the need for an external controller. This has been implemented through the use of a cross drive topology as shown in Figure 4 below. Both MOSFET switches are driven with the voltage across the opposite switch. This ensures that each MOSFET only switches when the voltage across it is close to zero. The nature of the resonance ensures that when one switch turns off, the other turns on. It is this alternation which allows for a full AC sinusoid to be produced at the resonant frequency. Figure 3 below depicts both the sinusoidal output from the resonant converter (across the track inductance) and the gate drive VDC S1 MCU Z Figure 4: Self sustaining resonant converter with low side switch IV. A. Pickup and compensation SECONDARY SIDE IMPLEMENTATION An inductor on the secondary (implant side) is required to pick up the high frequency magnetic fields produced by the primary resonant converter. When these magnetic fields couple with the secondary pickup inductor, they will induce a Z S2 CR LR

4 voltage by Faraday s law. This voltage will be utilized to charge the supercapacitor. The uncompensated pickup voltage can be represented with the equivalent circuit depicted in Figure 5 below. waveform. Although this arrangement is suitable for charging a supercapacitor, further modifications can be made. The addition of an inductor following the rectifier allows for the arrangement to mimic a DC current source. A DC current source is the most appropriate way to charge a supercapacitor. The basic charging arrangement is shown in Figure 6 below. I 1 L 2 Full-wave Rectifier L dc ω 2 M V oc R 2 L R2 C comp Figure 5: Uncompensated pickup circuit Where M is the mutual inductance. The induced open-circuit voltage in the secondary winding is: And the induced short-circuit current is: The maximum power output can be improved through the introduction of either parallel or series compensation [8]. Parallel compensation involves adding a capacitor in parallel with the pickup inductor, whilst series compensation involves adding a capacitor in series with the pickup inductor. (5) (6) Figure 6: Pickup compensation with additional charge circuitry V. SUPERCAPACITOR SUPPORT CIRCUITARY A. Active Balancing Circuit Present supercapacitor technology, although offering high capacitances, is limited by low voltage ratings. To overcome this, many manufacturers combine capacitor cells in series or parallel to increase the rated voltage or capacitance of the supercapacitor. In the present implementation, a similar approach has been adopted. The implemented charge cell configuration is shown in Figure 7 below. The maximum power for series and parallel compensation is the same, which can be expressed as [6]: (7) The distinction between the two forms of compensation comes from the fact that a parallel compensated pickup acts like an AC current source, whilst a series compensated circuit acts as an AC voltage source. Parallel compensation has been selected on the basis that a current source is more appropriate for capacitor charging applications [6]. The selection of the secondary inductor and capacitor depends on the resonant frequency of the primary converter: The values of and have been selected to achieve resonance at the operating frequency. The Q factor reflects the magnitude by which the compensated tank voltage has increased in relation to the open circuit uncompensated voltage. B. Optimizing pickup for supercapacitor charging The AC current source produced in a parallel compensation arrangement is unsuitable to charge a supercapacitor; a DC source is required. A full wave rectifier has been added after the compensation tank to produce a DC (8) Figure 7: Implemented supercapacitor bank As shown by Figure 7, the supercapacitor is composed of four individual supercapacitors. The series capacitances double the rated voltage to 5.5V, but the capacitance reduces to half. The parallel capacitances increase the overall capacitance back to. This configuration has allowed for an increase in rated voltage, without sacrifice of the total capacitance. Leakage current is a parasitic effect in all capacitors. It is often modeled as a resistance in parallel with the capacitor. Leakage current is generally negligible. However, when two capacitors are in series, an imbalance in the leakage current between the cells could result in one of the supercapacitors exceeding its rated voltage. Rudimentary solutions to this problem include placing identical bleed resistors across the series capacitances, these resistances are selected such that they draw an order of magnitude more than the leakage current [10].

5 Although being simple, this solution can result in a high leakage current (up to 50µA) from the supercapacitor, making this unsuitable for the intended application. To overcome the high leakage currents associated with placing bleed resistors across the series capacitances, an active balancing circuit has been implemented. The circuit topology is shown in Figure 8 below. R divide R divide + - Opamp Figure 8: Active charge balancing circuit It has been shown that the implementation of this balancing topology reduces the total leakage current by a factor of ten (to approximately 3.6 µa), whilst ensuring none of the capacitor elements are allowed to exceed their rated voltage [9]. B. DC-DC voltage regulation In a traditional battery operated system, as the battery discharges, its output voltage remains relatively constant. This means that in some situations a battery can be connected directly across the load it is driving. The behavior of supercapacitors however is significantly different. As a supercapacitor discharges, its voltage decreases. Although this makes it easy to determine the state of charge of a supercapacitor, additional circuitry is required to ensure it can adequately supply the load. Given the discharge characteristic of a supercapacitor, a DC-DC converter is required to ensure a constant output voltage. A Buck-Boost switching converter topology was initially employed to ensure constant output voltage. This switching converter topology is required given the wide voltage range of the supercapacitor. A problem with many of the Buck-Boost IC s explored was that their minimum input voltage was limited to approximately 1.8V. This means that when the supercapacitor terminal voltage discharged to below 1.8V, a constant output voltage could no longer be sustained. The remaining supercapacitor energy was effectively wasted. To overcome this problem, it was decided that the Buck-Boost switching converter would be placed in parallel with a Boost converter. Boost converters generally have a lower minimum input voltage rating generally about 0.6V. This means that, the capacitor can discharge for a longer period of time before the output voltage collapses. VI. DIGITAL CIRCUIT IMPLEMENTATION Wireless transceiver units were implemented on both the primary and secondary sides to allow for continuous data transmission from the sensor unit. Two Nordic NRF2401A RF transceivers were employed for this purpose. The wireless transceivers allow for periodic data transfer from the secondary implant to the primary side. A computer interface has been achieved through the use of the RS-232 protocol. The primary side transceiver picks up the wirelessly transmitted data, and transfers it to a computer via the RS-232 interface. A. Digital charge control Previous implementations of supercapacitor charge control circuitry rely on active switches to stop charging the capacitor when its voltage exceeds the rated value [7]. This approach however, is not suitable for implantable sensor applications. One of the primary considerations with any implantable electronic device is size. In order to minimize the associated risk of large implantable devices, PCB size should be kept to an absolute minimum. Power dissipation is also an important consideration with implantable devices. A traditional charge control topology would result in high power dissipation across the switching transistor, as it is required to pass all the charging current [7]. A digital charge control scheme has been implemented to regulate the supercapacitor charging. The wireless communication link has been utilized to allow for digital charge control. A block diagram of this charging configuration is shown in Figure 9 below. 1 o MCU Initiates 1 o Transceiver Voltage Reading Received Voltage UART to MatLab at 8N Baud 1 o MCU Starts Resonant Converter Charging Voltage Reading Transmitted Voltage above Upper Threshold? No Voltage below Lower Threshold? Yes Yes No System Turned On DC-DC Converter starts 2 o MCU Initiates 2 o Transceiver ADC of Voltage 1 o MCU turns Resonant Converter Off 1 o MCU turns Resonant Converter On Figure 9: Block diagram of implemented digital charge control This implementation relies on the low side switch on the resonant converter. This was discussed previously in 3. When charging initially begins, the low side switch is driven on. This allows the supercapacitor to charge. Once the supercapacitor has charged to approximately 0.8V, the secondary

6 microcontroller and wireless transceiver will turn on. From this point the microcontroller will continually sample the value of the supercapacitor terminal voltage and transmit it to the primary microcontroller. If the voltage exceeds a predefined limit (was set as 5.5V as this is the rated voltage of the selected capacitor), the microcontroller will switch off the low side switch on the resonant converter. This will immediately cease the supercapacitor charging. As previously explained, the data transmission continues as the supercapacitor discharges. When the capacitor has discharged below a predefined lower threshold (set at 1V), the resonant converter will switch back on, thus recharging the supercapacitor. Figure 10 shown below depicts the supercapacitor charging waveform. As shown, when the supercapacitor reaches the upper voltage threshold, charging ceases. The charge waveform indicates that the supercapacitor (, 5.5V) can be charged in approximately 5 seconds. This charging system has been integrated with a digital charge control system which allows for supercapacitor charging to be ceased using a microcontroller, and wireless communication link between the primary and implant side. This communication link also allows for real time data transfer between the implanted module and an external device. To ensure supercapacitor energy utilization is maximized, a hybrid DC-DC converter has been implemented. This hybrid topology involves placing a Buck-Boost converter in parallel with a Boost converter. Through the implementation of this hybrid DC-DC converter topology, it was found that the supercapacitor could sustain circuit operation for approximately 60 minutes with a load of about 6mW. Presently, limitations in supercapacitor capacity restrict its wide acceptance in replacing batteries. As the capacities of supercapacitors increase, this disparity will diminish and it is predicted that supercapacitors will become more common substitutes for traditional batteries. REFERENCES [1] CAP-XX (Australia) Pty Ltd, Power vs. Energy, [Online] Available: [Accessed: May 3, 2010]. [2] Brunelli, D.; Moser, C.; Thiele, L.; Benini, L.;, "Design of a Solar- Harvesting Circuit for Batteryless Embedded Systems," Circuits and Systems I: Regular Papers, IEEE Transactions on, vol.56, no.11, pp , Nov [3] Simjee, F.; Chou, P.H.;, "Everlast: Long-life, -operated Wireless Sensor Node," Low Power Electronics and Design, ISLPED'06. Proceedings of the 2006 International Symposium on, vol., no., pp , 4-6 Oct Figure 10: charging waveform The explanation described above assumes that the supercapacitor voltage is the only data transmitted. It is more likely that the supercapacitor voltage will be transferred in conjunction with some other information (such as an ECG signal). In this case, an extra bit will be transmitted indicating the type of data. This would allow for the implementation of both digital charge control and data transmission of an additional sensed signal. VII. CONCLUSIONS This paper has investigated the potential for supercapacitors to replace batteries in low power implantable sensor environments. Initial feasibility analysis showed supercapacitors are ideal to replace batteries in low power applications when fast charging and discharging are required. In these applications, the high power density of the supercapacitor is exploited to significantly decrease the charge time. Specifically, an IPT based supercapacitor charging system has been developed. This system allows for a supercapacitor to be fully charged in approximately 5 seconds. [4] Telemetry Research Ltd, Charging of Transmitters, [Online]. Available: =view&id=39&itemid=164. [Accessed: May ] [5] CAP-XX (Australia) Pty Ltd, Battery Types and Uses, [Online] Available: [Accessed: May 3, 2010]. [6] Hu, A.P: Selected resonant converters for IPT power supplies, PhD thesis, Department of Electrical and Computer Engineering, University of Auckland, Oct [7] Abeywardana, D K., Wireless power supply for low power smart sediment particle, Masters Thesis, Department of Electrical and Computer Engineering, University of Auckland, July [8] Hu, A.P., Kwan, I.L.W., Tan, C, and Li.Y, A wireless battery-less computer mouse with supercapacitor energy buffer, in ICIEA nd IEEE Conference on Industrial Electronics and Applications. May [9] Boys, J.T., Covic, G.A. and Green, A.W.: Stability and control of inductively coupled power transfer systems, IEE Proceedings on Electric Power Applications, vol.147, pp 37-43, Jan [10] CAP-XX Ltd., DRAFT: Active Voltage Balancing for s, CAP-XX Ltd., V.10, pp.2, May 2009.,