A Low-Power Sensor Read-Out Circuit with FSK Telemetry for Inductively-Powered Implant System

M. R. Haider+, S. Mostafa and S. K. Islam

Department of Electrical Engineering and Computer Science, The University of Tennessee Knoxville, TN 37996-2100 E-mail: [email protected]
AbstractThis paper presents a low-power sensor read-out circuit with FSK telemetry option. The proposed system consists of a data generator block and a FSK generator block. The data generator block converts a sensor current to a square wave data signal whereas the FSK generator block produces two different frequencies depending upon the value of the data signal. The relatively simple architecture of the proposed system facilitates the low-voltage and low-power operation. The system has been designed using AMI 0.5 m CMOS process. Simulation results show that the proposed system can detect a sensor current in the range of 50 nA to 5 A and the total power consumption is 431 W with a 3.3 V supply. The circuit is ready to be submitted for monolithic integration.

attractive solution for a number of issues, such as wireless powering and data communication, cable free operation with reduced risk of infections, increased implant robustness and sufficient miniaturization. The maximum achieved efficiency for transdermally-provided power to biomedical implants, as reported by Djemouai and Sawan, is below 20% [2], and it could be as low as 0.25% [3] or even lower. Monolithic integration of sensors and signal processing unit offers a power and area efficient design of the implanted system. Johannessen et al have reported a microelectronic pill fabricated in a 0.6 m CMOS process that consumes 12.1 mW of power, measures 5.51.6 cm2, and weighs 13.5 gm including two silver-oxide batteries [4]. Recently Mohseni et al have developed miniature wireless frequency-modulated systems for biopotential recording and telemetry applications with substantially lower power dissipation per number of recording channels. The total chip area is 4.84 mm2, and the power consumption is 2.2 mW with a 3 V power supply [5]. Most of these approaches are complex, power-hungry and occupies a large area. Recently a board-level implementation of inductive power link has been demonstrated [6] which has a power transfer capability of 125 mW with 5V power supply. The prototype inductive link uses FSK modulation scheme to transmit power to the receiving unit. Since the future goal is to transmit power and data through the same coil, FSK scheme has been used for both data and power transmission to avoid any power amplifier nonlinearity issues as well as to achieve better noise immunity. Again, the sensor signal-processing unit should be capable of working with the received power that necessitates the design of low-voltage low-power signal-processing unit. Therefore, in this paper we are proposing a low-voltage lowpower sensor read-out circuit with FSK telemetry option for implantable sensor applications. The entire system consists of a data generator block and a FSK generator block. Simulation results show that the entire system consumes only 431 W of power with a 3.3 V power supply.

I.

INTRODUCTION

With the availability of micro- and nano- fabrication processes and wireless communication technologies, recent technological improvements of healthcare monitoring equipments, have led to the development of miniature, lightweight, and energy-efficient circuit solution for biomedical sensor applications. In most cases, the sensor output is an analog current signal, which needs to be further processed and transmitted out of the biological environment (i.e. human body) so that the signal can be detected externally for analysis and initiation of necessary actions. The processing can include such important features as self-calibration, self-diagnostics, and advanced pattern recognition. All these functional design elements can be enclosed in the sensor package that provides measurement integrity to the device. For any biomedical application, implantable sensors have two common requirements: i) a means for non-invasive, reliable and efficient power supply to the implanted unit, and ii) a means for robust and reliable data communication between the implanted unit and an external unit. Various methods have been reported in the previous years to power up the implants but none of them offers risk free operation. Such methods include transcutaneous power cables [1] which introduce a significant path for infection and usage of batteries introduces a potential risk of leakage. An inductive link does not have any of these limitations, and thus offers a very

Recovered Data

Demodulator Unit

Data

Power Amplifier
Power

FSK Generator Block

Data

External Power Unit

Power External Coil Internal Coil

Power Recovery Unit

Data Generator Block

Sensor Biasing

Skin
Figure 1: Inductively-powered implant system

II.

SYSTEM ARCHITECTURE

following paragraphs.

Fig 1 shows the block diagram of an inductively-powered implant system. The entire system manifests external power unit, link coils, power recovery unit, data generator block, FSK generator block, power amplifier and data demodulator unit. The external power unit energizes the external coil to inductively couple power to the internal coil that resides underneath the skin. The power recovery unit works in resonance condition to convert the ac-coupled power into a dc voltage and feds this voltage to the rest of the implant system to perform their functionality. The data generator block generates a pulse width modulated signal based on the sensor current. This data signal is then fed into the FSK generator block where two different frequencies are generated based on the High and Low level of the data signal. The FSK signal is then sent to a power amplifier to transmit the data using the same link coils. The demodulator block extracts the data signal from the external coil and makes the data usable for further processing and diagnostics. In this paper we are focusing on the data generator block and FSK generator block only and the detail descriptions of these two blocks are given in the

Data Generator Block The data generator block as shown in fig 2 is fed a sensor current that can be generated by any implantable electrochemical sensor. A current mirror (N1, N2) is used at the input part of the data generator circuit. The mirror current charges the integrating capacitor (Cint) to the supply voltage VDD resulting in a decreasing voltage at the input of the Schmitt trigger. At the moment, the voltage goes below the threshold voltage of the Schmitt trigger the PMOS switch turns on and discharges the capacitor to start the new cycle. The charging rate of the capacitor is controlled by the mirror current which is in turn proportional to the sensor current. Thus the oscillation frequency of the data signal is directly proportional to the sensor current level. An additional inverter (inverter 2) is used to minimize the loading effect on data signal. Finally, a D-latch based frequency divider is used at the last stage of data generator block to make the duty cycle of the data signal to almost 50% and to provide at least ten times frequency separation between the data and the lowest carrier frequency of the FSK signal.

A.

Figure 2: Data Generator Block Figure 3: FSK Generator Block

equivalent capacitance value can be expressed as

Ceq , Low = C p +

C1 C2 C1 + C2

(1)

When the data signal is High, Md shortens the C2 capacitor and the equivalent capacitance value can be expressed as

Ceq , High = C p + C1

(2)

Figure 4: Data signal for sensor currents of 2 A (top trace) and 50 nA (bottom trace)

Now from equation (1) and (2), it is obvious that whenever the data signal is Low, the equivalent capacitance value is lower due to series combination of C1 and C2 and the output frequency is higher. On the other hand, whenever the data signal is High, the equivalent capacitance value is higher and ultimately the output frequency is higher. Thus the FSK generator block is capable of generating two different frequencies depending upon the value of data signal.

B. FSK Generator Block The detailed block diagram of the FSK generator block is shown in fig 3. It consists of a Schmitt trigger, an inverter, a switched-capacitor structure and a NMOS (M1) to periodically discharge the capacitor. The beta-multiplier generates a constant current without process, voltage and temperature (PVT) variations. This constant current charges the switchedcapacitor structure towards the supply voltage. At the moment, the capacitor voltage reaches the upper threshold voltage of the Schmitt trigger, the Schmitt trigger generates a negative pulse which is then inverted by the inverter and a positive voltage is fed back to the NMOS (M1) to discharge the capacitor. The discharge rate depends upon the value of capacitance and on the equivalent resistance of M1. To achieve 50% duty cycle of the carrier frequency, the on resistance of the M1 is optimized with the equivalent value of the switchedcapacitor structure. Equivalent higher resistance is obtained by increasing the length of NMOS. The switched-capacitor structure consists of series-parallel combination of C1, C2 and Cp. An NMOS (Md) is placed in parallel with the C2 capacitor and the gate of this MOSFET is driven by the data signal. When the data signal is Low, the

III.

SIMULATION RESULTS

The entire system has been designed and simulated with AMI 0.5 m bulk CMOS process. Cadence SpectreTM simulator has been used to check the performance of the proposed system. The relatively simple architecture and reduced number of transistors make the design highly suitable for low-voltage low-power operation. For simulation purpose, an ideal current source is used to mimic the behavior of any electrochemical sensor under varying conditions. Sensor biasing is assumed to be 0.7 V [7] and a 400 fF capacitor is added to the drain terminal of N1 for the inclusion of pad parasitics. Fig 4 shows

Figure 5: Data frequency variation for different sensor currents

Figure 6: Data (top trace), FSK (middle trace) and frequency variation of FSK (bottom trace) signal for a sensor current of 1 A

the generated data signal for a sensor current of 50 nA and 2 A for a supply voltage of 3.3 V. Fig 5 shows a plot of the data frequency for different sensor current levels. From fig 5, it is evident that the data generator block can generate a linear frequency variation for sensor currents in the range of 50 nA to 5 A. Fig 6 shows the data and FSK signal for a sensor current of 1 A with a 3.3 V supply voltage. The bottom trace of fig 6 indicates the high and low carrier frequency to be 1.0159 MHz and 536 KHz, respectively. By proper setting of frequency divider, sufficient frequency between the lowest carrier frequency and the data signal can be easily made.
TABLE I: TOTAL POWER CONSUMPTION WITH DIFFERENT SENSOR CURRENTS

TABLE III: DEVICE DIMENSIONS OF NMOS AND PMOS

Device N1, N2 PMOS switch Md M1 IV.

Dimension 1.5m/1.2 m 1.5m/1.05 m 1.5m/1.2 m 1.5m/12 m

CONCLUSION AND FUTURE WORKS

Sensor Current (A) 0.05 0.5 1 1.5 2 3 4 5

Power (W) 429 412 431 436 447 480 536 626

In this paper we have proposed a low-power sensor readout circuit with FSK telemetry scheme for low-power implant system. The data generator circuit is capable of sensing sensor current in the range of 50 nA to 5 A. The FSK generator circuit can generate two distinct frequencies depending upon the signal amplitude of the data signal. The entire system can operate with a supply voltage as low as 3.3 V and consumes only 431 W of power for a sensor current of 1 A. Relatively simple architecture and low-voltage low-power operation make the proposed scheme a potential candidate for its applications in inductively-powered implant system. REFERENCES
[1] D. C. Galbraith, M. Soma, and R. L. White, A wideband efficient inductive transdermal power and data link with coupling insensitive gain, IEEE Transactions on Biomedical Engineering, vol. BME-34, pp. 265-275, April 1987. A. Djemouai, and M. Sawan, Wiley Encyclopedia of Electrical and Electronics Engineering, vol. 17, pp. 413, 1999. R. Puers, M. Catrysse, G. Vandevoorde, R. J. Collier, E. Louridas, F. Burny, M. Donkerwolcke, and F. Moulart, A telemetry system for the detection of hip prosthesis loosening by vibration analysis, Sensors and Actuators, vol. 85, pp. 42-47, 2000. E. A. Johannessen, L. Wang, L. Cui, T. Tang, M. Ahmadian, A. Astaras, S. Reid, P. Yam, A. Murray, B. Flynn, S. Beaumont, D. Cumming, and J. Cooper, Implementation of multichannel sensors for remote biomedical measurements in a microsystem formats, IEEE Transactions on Biomedical Engineering, vol. 51, no. 3, pp. 525-535, March 2004. P. Mohseni, K. Najafi, S. J. Eliades, and X. Wang, Wireless multichannel biopotential recording using an integrated FM telemetry circuit, IEEE Transactions on Neural Systems and Rehabilitation Engineering, vol. 13, no. 3, pp. 263-271, September 2005. M. R. Haider, M. A. Huque, M. A. Adeeb, S. K. Islam, W. Qu, and N. Ericson, A low-power sensor-signal read-out circuit powered by inductive link, 8th International Conference on Solid-State and Integrated-Circuit Technology, pp. 1643-1645, October 23-26, Shanghai, 2006. M. R. Haider, S. K. Islam, and M. Zhang, A low-power signalprocessing unit for in vivo monitoring and transmission of sensor signal, Sensors & Transducers, vol. 84, issue 10, pp. 1625-1632, October 2007.

[2] [3]

Table I depicts the total power consumption for different sensor current levels. The lowest and highest power consumptions within the sensing range are found to be only 429 W and 629 W, respectively. Due to this low-power
TABLE II: CAPACITANCE VALUES FOR DATA AND FSK GENERATOR BLOCKS

[4]

Component Cint C1 C2 Cp

Value 5 pF 2.4 pF 0.8 pF 1.2 pF

[5]

[6]

consumption, the proposed system is highly suitable for any low-power sensor applications. Table II and III have shown the capacitance values and device dimensions of data and FSK generator blocks.

[7]