A 2.4W 400nC/s Constant Charge Injector for Wirelessly-Powered Electro-Acupuncture

Hyungwoo Lee, Kiseok Song, Long Yan, and Hoi-Jun Yoo Department of Electrical Engineering

Korea Advanced Institute of Science and Technology (KAIST) Daejeon, Republic of Korea

hwlee@eeinfo.kaist.ac.kr

Abstract—An ultra-low-power constant charge injector (CCI) circuit is presented for wirelessly-powered electro-acupuncture (EA). The CCI adopts adaptive pulse-width and frequency-drift (APF) calibration loop that not only accommodates to body impedance variation (BIV) of 100-200k but also is tolerable to frequency-drift. The proposed 5Hz current starving clock generator combined with sub-Vth reference circuit provides supply voltage dependency of 0.2Hz/V and temperature dependency of 0.018Hz/oC while consuming only 1W. These variations are finely calibrated by the APF loop to ensure inject-able charge intensity as stable as 399.33-400.45nC/s. The low-power low-voltage Gm circuit in CCI incorporates diode-limited and chopper-modulated inputs to prevent the damage from static electricity of the body while providing linear voltage-to-current conversion. The proposed CCI, simulated in 0.18m CMOS technology with supply voltage of 1.0V, consumes 2.4W of power.

I. INTRODUCTION

An invasive medical treatment using electro-acupuncture (EA) [1] is a promising research area, and it has received more and more attention on the effectiveness for myalgia [2], depression [3], and anesthesia [4]. A pair of needles is exploited in EA to inject charge into the body by flowing electrical current from a bulky and wire-connected power supply.

To improve remedial value and subject’s convenience, a multi-channel wirelessly-powered EA system [5], [6] was proposed to remove cumbersome wires and to give the EA a compact form. To cope with limited power budget (<8W), low supply voltage (1.0V), and injecting constant charge issue with respect to body impedance variation (BIV) of 100-200k, it adopts an adaptive pulse-width (APW) control loop in which a chopper modulated Gm circuit with flipped voltage follower (FVF) is integrated. However, the previous approach [5], [6] of Fig. 1 still remains several technical issues to be answered. First, although the APW stimulator can detect the

BIV dynamically, the voltage change across the BIV is not protected to be vulnerable to static electricity. Therefore, it may damage the succeeding Gm circuit so that the APW fails to proper operation. Second, the separated implementation between the reference circuit and the ring oscillator in APW will introduce large frequency-drift and power consumption. Third, the safety issue correlated to balanced charge intensity [7], which requires exact stimulation pulse frequency, was not considered.

In this paper, a hybrid frequency-drift compensation mechanism is presented for adaptively controlling stimulation pulse-width so that injected charge should be stable over time. The drift of the stimulation frequency as the functions of supply voltage and temperature is coarsely controlled to be a minimum by a low-power constant current starving ring oscillator combined with sub-Vth reference generator. It was finely regulated by adaptive pulse-width and frequency-drift (APF) calibration loop. To be resilient to static electricity [8], [9] and to ensure reliable operation of the APF loop, a diode-limited and Gm circuit incorporating chopper modulation is integrated.

This paper is organized as follows. Section II describes the system design challenges and operation of the proposed APF loop. Section III shows low-power sub-Vth reference and clock

Figure 1. The previous wirelessly-powered EA [5], [6]

978-1-4244-9472-9/11/$26.00 ©2011 IEEE 1716

frequency generator in detail. After the simulation results shown in Section IV, Section V will conclude the paper.

II. THE PROPOSED CONSTANT CHARGE INJECTOR BASED

ON ADAPTIVE PULSE-WIDTH AND FREQUENCY-DRIFT

CALIBRATION LOOP

Fig.2 shows the structure of the proposed constant charge injector (CCI) based on APF calibration loop. It consists of: 1) diode-limited protection circuit prior to Gm-stage so that it is isolated from static electricity, 2) a low-power low-voltage chopper-modulated Gm circuit [6], which linearly translates the voltage of the BIV into current signal regardless of process mismatch, 3) hybrid frequency-drift compensation blocks including low-power coarse frequency-drift controllable reference and clock generator together with fine frequency-drift constant current drain circuit, and 4) auxiliary control logic and EA driver for injecting charge. The operation of the proposed APF compensation loop is divided into 5 steps.

Step-1: Once BIV (100-200k) is introduced, the integrated BIV sensing resistor of RO (10k) converts it into 45-95mV voltage signal (VRO).

Step-2: The large static electricity superimposed to VRO will be by passed to ground via diode-limiter circuit, and only VRO is fed into Gm-stage for translation into current signal of ICS to charge CS in Fig.2.

Step-3: ICS flowing into CS will increase the VCS of Fig.2 when clock signal (CLK) is high. Once the VCS equals to VREF of Fig. 2, the charging switch of SW1 will be turned off to determine the stimulation pulse-width of tpw in (1). As such, the inject-able amount of charge (Qinj) is calculated as (2).

DDO

BIV

m

REFSpw VR

R

G

VCt

(1)

Om

REFSinj RG

VCQ

(2)

Step-4: When CLK is low, the switch of SW2 turned on to discharge CS with a constant current of IB. It results in voltage drop of ΔVCS in (3) which is inversely proportional to the frequency of CLK (fCLK).

CLKS

BCS fC

IV

2 (3)

Step-5: In the succeeding CLK period, the Step-3 and Step-4 are repeated to correct tpw as (4). As such, the inject-able amount of charge (Qinj) will be modified as (5).

CLK

B

DDOm

BIVpw f

I

VRG

Rt

2 (4)

Figure 2. The proposed constant charge injector (CCI) incorporates adaptive pulse-width frequency-drift calibration and ESD protection

Figure 3. Timing diagram of the APF calibration loop Figure 4. Chopper-modulated Gm circuit [6]

1717

CLKOm

Binj fRG

IQ

1

2

(5)

Consequently, the inject-able charge intensity (Qinjtpw) of Fig.3 becomes constant regardless of drift of fCLK and BIV as (6), but depends on constant value of current drain, Gm, and RO. To minimize the contribution from non-ideality of Gm, the chopper-modulated Gm circuit [6] of Fig.4 is adopted to ensure precise linearity (0.2A/V) with respects to process mismatch of (10%), which affects only 2% of tpw while consuming only 1W.

Om

BCLKinj RG

IfQ

2 (6)

III. LOW-POWER SUB-VTH REFERENCE AND CLOCK

FREQUENCY GENERATOR

To achieve a constant IB in (6) with low-power consumption, a sub-Vth constant current generator of Fig. 5 (a) is utilized [10].

Fig. 5 (a) shows the schematic of the sub-Vth reference and clock frequency generator. In order to implement with a high tolerance to a frequency-drift due to supply voltage variation and temperature variation, we use two schemes: First, we compensate the temperature dependency with a combination of complementary to absolute temperature (CTAT) and proportional to absolute temperature (PTAT) circuit. When temperature increases, CTAT circuit outputs the linearly decreased current. However, the PTAT circuit has positive slopes as the function of temperature variation with increased output current. The ratio of increasing or decreasing currents can be adjusted so as to sum the two slopes up to be zero. At the same time, output current dependency to the variation of supply voltage is coarsely regulated PTAT current biased structure. The ring oscillator driven by currents depends entirely on the amount of current, not by other environments. As such, we can extract not only the accurate reference current but also the precise reference voltage.

Fig. 5 (b) shows the stability of the reference current with respect to the variation of the temperature and supply voltage. This reference current is copied to provide IB of 1.9nA and VREF of 0.6V. In the case of reference current, it shows only 1.57% variation when temperature is varied from 0 to 60oC. Moreover, it has only 1.49% variation when the supply voltage has been changed from 0.9 to 1.1V. The results are sufficiently stable to ensure APF operation.

IV. SIMULATION RESULTS

Fig. 6 shows the simulation results of the proposed APF calibration loop. The APF with the fCLK of 4Hz provides the constant Qinj of 98nC/stimulation, which is shown in (2), with respects to tpw of 10.2-20.0ms and BIV of 100-200k. As the fCLK is increased to 6Hz, the Qinj is decreased to 66.8nC/stimulation, which well verifies (5), with respects to tpw of 13.62ms and the BIV of 200k. In both cases, the inject-able charge intensity maintains constant of 400nC/s which also verifies (6).

Fig. 7 shows the stability of the inject-able charge intensity as the function of (a) supply voltage variation and (b)

Figure 6. Simulated waveforms of the APF calibration loop

(a)

(b)

Figure 5. (a) Schematic and (b) simulation results of the sub-Vth reference and clock frequency generator

1718

temperature variation. The sub-Vth reference and clock frequency generator provides the sensitivity of 0.2Hz/V and

0.018Hz/℃, respectively. As a result, the stimulation frequency drifts only between 4.995Hz and 5.035Hz over supply voltage variation of 0.9-1.1V, and between 4.64Hz and

5.7Hz over temperature variation of 0-60℃, respectively. Within this frequency range, the APF stabilizes the inject-able charge intensity of 399.33-400.45nC/s with the variation of 0.4nC/s (0.01%) and 1.12nC/s (0.3%), respectively.

V. CONCLUSION

A 2.4W CCI is presented for wirelessly-powered EA. The CCI incorporates APF calibration loop to finely calibrate stimulation pulse-width for maintaining constant charge intensity regardless of BIV (100-200ksupply voltage variation (0.9-1.1V), and temperature variation (0-60oC). The simulation results, under 0.18m CMOS technology with supply voltage of 1.0V, show that the proposed APF in CCI provides supply voltage dependency of 0.2Hz/V and temperature dependency of 0.018Hz/oC, which is sufficiently

stable for inject-able charge intensity of 399.33-400.45nC/s. For the low-power consumption of CCI, the ring oscillator is embedded in reference circuit operated in sub-Vth, and its frequency is coarsely controlled by a constant current starving oscillator. The proposed CCI prevents the damage from the static electricity of the body with a diode-limiter in APF, and it is expected to be more safe and lower power consumption than the previous APW stimulator [5],[6].

REFERENCES [1] T. J. Kaptchuk,”Acupuncture: Theory, efficacy, and practice,”

Academia and Clinic, Annals of Internal Medicine, vol.136, no.5, pp.374-383, March 2002.

[2] M. Sandberg, B. Larsson, L. G. Lindberg and B. Gerdle,“Different patterns of blood flow response in the trapezius muscle following needle stimulation (acupuncture) between healthy subjects and patients with fibromyalgia and work-related trapezius myalgia,” European Journal of Pain, vol.9, pp.497-510, December 2004.

[3] H. Luo, F. Meng, Y. Jia and X. Zhao,“Clinical research on the therapeutic effect of the electro-acupuncture treatment in patients with depression,” Psychiatry and Clinical Neurosciences, vol.52, pp.338-340, 1998.

[4] J. Lin, M. Lo, Y. Wen, C. Hsieh, S. Tsai and W. Sun,“The effect of high and low frequency electroacupuncture in pain after lower abdominal surgery,” Pain, vol.99, issue.3, pp.509-514. October 2002.

[5] K. Song, S. Lee, and H.-J. Yoo,“A wirelessly-powered electro-acupuncture based on adaptive pulse width mono-phase stimulation,” IEEE International Symposium on Circuits and Systems, pp.2087-2090, May 2010.

[6] K. Song, L. Yan, S. Lee, J. Yoo, and H.-J. Yoo, “A Wirelessly-Powered Electro-Acupuncture based on Adaptive Pulse Width Mono-Phase Stimulation,” invited to IEEE Transactions on Biomedical Circuits and Systems (in review progress) , special issues on IEEE International Symposium on Circuits and Systems 2010.

[7] C. David Lytle, Brenton M. Thomas, Edward A. Gordon and Victor Krauthamer.,“Electrostimulators for acupuncture: safety issues”, The Journal of Alternative and Complementary medicine, vol.6, 2000.

[8] A. Amerasekera and W. Verwey,“ESD in integrated circuits,” Quality and Reliability Engineering International, vol.8, no.3, pp.249-272, 1992.

[9] R. Merrill and E. Issaq,“ESD Design methodology,” the Electrical Overstress/Electrostatic Discharge Symposium, pp.223-237, September 1933.

[10] N. Cho, S.-J. Song, J.-Y. Lee, S. Kim, S. Kim, and H.-J. Yoo,“A 5.1-uW UHF RFID Tag Chip integrated with Sensors for Wireless Environmental Monitoring,” Proc. of IEEE European Solid-State Circuits Conference (ESSCIRC), pp.279-282, 2005.

(a)

(b) Figure 7. The stability of inject-able charge intensity as the function of (a) supply voltage variation and (b) temperature

variation

1719