High-speed OQPSK and efficient power transfer through inductive link for biomedical implants

192 IEEE TRANSACTIONS ON BIOMEDICAL CIRCUITS AND SYSTEMS, VOL. 4, NO. 3, JUNE 2010

High-Speed OQPSK and Efficient Power TransferThrough Inductive Link for Biomedical Implants

Guillaume Simard, Student Member, IEEE, Mohamad Sawan, Fellow, IEEE, andDaniel Massicotte, Senior Member, IEEE

Abstract—Biomedical implants require wireless power and bidi-rectional data transfer. We pursue our previous work on a noveltopology for a multiple carrier inductive link by presenting thefabricated coils. We show that the coplanar geometry approach isbetter suited for displacement tolerance. We provide a theoreticalanalysis of the efficiency of power transfer and phase-shift-keyingcommunications through an inductive link. An efficiency of up to61% has been achieved experimentally for power transfer and adata rate of 4.16 Mb/s with a bit-error rate of less than 2 10 � hasbeen obtained with our fabricated offset quadrature phase-shiftkeying modules due to the inductive link optimization presentedin this paper.

Index Terms—Biomedical communication, biomedical implant,inductive link, power transfer, quadrature amplitude modulation.

I. INTRODUCTION

C LASSICALLY, biomedical implants depending on induc-tive links for power and bidirectionnal data transfer have

been designed with only one link. This link had to be shared tomeet often contradictory requirements and a general compro-mise had to be made. A typical example is bandwidth whichneeds to be widened for data transfer but narrowed for efficientpower delivery. Back telemetry could also be achieved usingload-shift keying (LSK), which allows the external controller tosense variations of its coil current. This modulation scheme isslow (100 kb/s) and problematic with complex implants wherethe overall current is not constant.

Within recent years, people have begun to push the down-link data rates above the 2.5-Mb/s [1] threshold. Back telemetry,however, attracted much less attention from researchers so far,until they introduced the possibility of using RF for this pur-pose [1], making their system a hybrid RF and dual-band induc-tive link. Others are proposing dual-band inductive links [2], [3],keeping load-shift keying (LSK) as a possible feedback path. Wefollow this avenue and separate power from data, the later beingitself split into two more links to allow for a good crosstalk-free

Manuscript received September 16, 2009; revised November 17, 2009. Firstpublished February 05, 2010; current version published May 26, 2010. Thiswork was supported in part by the Canada Research Chair on Smart Medical De-vices, in part by the NSERC research grants, in part by an NSERC ES M Grad-uate Scholarships Award, and in part by an ReSMiQ Scholarship. This paperwas recommended by Associate Editor T. Constandinou.

G. Simard and M. Sawan are with the Polystim Neurotechnologies Labora-tory, Ecole Polytechnique de Montreal, Montreal, QC H3T IJ4, Canada (e-mail:[email protected]).

D. Massicotte is with the Department of Electrical Engineering, Universityof Quebec at Trois-Rivières, Trois-Rivières, QC G9A 5H7, Canada.

Color versions of one or more of the figures in this paper are available onlineat http://ieeexplore.ieee.org.

Digital Object Identifier 10.1109/TBCAS.2009.2039212

full-duplex bidirectional transfer. The novelty is that these threelinks are inductive in nature and independent from each other.

Although [1]–[3] present the orthogonal geometry for theirdual-band inductive link, none have made an attempt at back-telemetry through a third inductive link. The coaxial geometry,which is a special case of the coplanar geometry where coilsshare a common centroid, is also presented by [2]. In the nextsection, we compare the orthogonal geometry with our proposedcoplanar geometry and show the latter is better for cortical im-plants.

Polystim Neurotechnologies Laboratory’s work has pro-duced a high-speed bidirectional transceiver using the offsetquadrature phase-shift keying (OQPSK) modulation schemethat can be used for bidirectional transmission of information[4], [5]. The chip has been successfully tested experimentallythrough our new multiple carrier inductive link presented in[6]. See (1) at the bottom of the next page.

Fig. 1 shows the block diagram of our system. The doublelayer between each pair’s coils represents skin and is not to beconfused with the symbol for an iron-core transformer. Coils1–2 form the power transfer pair, coils 3–4 form the downlinkdata path (outside to inside), and coils 5–6 form the uplink datapath (inside to outside). The external controller is responsiblefor processing data from the implant, supplying power and pre-processed data from the outside world (i.e., sound, video, etc.).phase-shift keying (PSK)-based modulators and demodulatorsallow the internal control unit to communicate sensor readingsand activate electrical stimulators on demand, giving extra flex-ibility to the system. Our main contributions to this paper are asfollows.

1) We propose an efficient power transfer and higher full-du-plex data rate allowed by the minimization of crosstalk, aproblem present in most multiple carrier-based systems.

2) We compare the orthogonal geometry with our proposedcoplanar geometry and show the latter is better for corticalimplants.

3) We report the characterization of our fabricated coils, theefficiency of the inductive link for power transfer, and astrong improvement in the data rate of our OQPSK moduleallowed by our theoretical analysis of the propagation ofphase shifts through an inductive link.

We first present the topology of our inductive link and the-oretical predictions on its sensitivity to lateral displacements inSection III. Section II develops the theory behind inductive linksin general and then targets the efficiency of power transfer andpresents guidance to increased speed in the PSK data links. Fi-nally, we present experimental results comparing theory to mea-surements in Section V.

1932-4545/$26.00 © 2010 IEEE

SIMARD et al.: HIGH-SPEED OQPSK AND EFFICIENT POWER TRANSFER 193

Fig. 1. Block diagram of an implantable electronic device.

II. MAGNETIC INDUCTIVE LINK

A. Rationale for Biomedical Implants

Implants cannot extract sufficient energy from body heat orfrom nutrients within blood. An external source of electricalpower is necessary for these devices. Although percutaneouswires provide the optimal solution for power transfer, they arenot medically safe. Furthermore, an implanted battery wouldrequire repeated recharge or surgeries. This leaves us with alimited number of options, among which capacitive, magnetic,electromagnetic, optical, and magnetoelectric power transfer[7]. Mainstream research has shown that a magnetic inductivelink is well suited to the task of supplying power to biomedicalimplants and is used in many commercially available devicessince the 1960s. This paper proposes an efficient power transferand higher full-duplex data rate allowed by the minimizationof crosstalk, a problem present in most multiple carrier-basedsystems.

B. Air-Core Transformer Fundamentals

The well-developed theory of electrical circuits can be usedto analyze the magnetic coupling of two coils. In particular, ifthe operation frequency is low enough to use the lumped ap-proximation, the geometry of these coils can be simplified tothree parameters: their respective self inductances andand their mutual inductance . The level of interaction be-tween two coils and can be appreciated by comparing it totheir maximum possible interaction. This quantity is the cou-pling coefficient [1]

(2)

Fig. 2. Schematic of the inductive link used for power transfer.

Equation (1), shown at the bottom of the page and derivedfrom Fig. 2, shows gain which can be made maximal andsimplified extensively if the inductors (L) and capacitors (C)of the tanks are chosen so that their natural resonance frequen-cies are the same and correspond to the inputsignal’s frequency . The differentiation of with re-spect to provides the critical coupling coefficient whichmaximizes gain [1]. If is neglected and ,is given by

(3)

This link transfers maximimum power when ismaximum. Matching primary resistance with the reflectedsecondary resistance and cancelling reactive components pro-vide this maximum. These conditions are known as the max-imum power transfer theorem. Equation (4) gives the impedanceseen at source and shows how secondary’s impedance istransformed when viewed from the primary side

(4)

Maximizing power transfer (equivalent toand ) implies working with an efficiency50% which is by all means unacceptable in biomedical implants.The external battery would deplete much faster, and excess heatwould be absorbed by the body.

C. Maximizing Efficiency of Power Transfer

By definition, efficiency is the ratio of power consumed bythe load to the power delivered by the source [8]

(5)

The full expression for efficiency is derived from (5) andshown in (6). The denominator is composed of two parts: 1)corresponding to the efficiency of power transfer through the

(1)


Fig. 3. Secondary equivalent serial transformation.

link and 2) the efficiency of the secondary side of the link,respectively

(6)

In deriving this equation, the secondary side of the circuithas been transformed by using (7)–(9) as illustrated in Fig. 3.The serial transformation, instead of a parallel transformation,makes the equations simpler to handle

(7)

(8)

(9)

It is important to note that the transformation embodied in (8)and (9) does not introduce the narrowband approximation usedin many papers on the subject.

It is quite obvious that and have a central role inthe upper bound of . The ratio of the secondary circuitalso has an impact on efficiency. Given a fixed operation fre-quency, one should strive to minimize the ratio while main-taining the quality factor of the coils because in an oscillatingLC tank . Losses are essentially a function of

while power dissipated in the load depends on. The value of the capacitor at the primary side has, how-

ever, no influence on efficiency except for its equivalent serialresistance. Gain may then be adjusted by varying withoutincidence on efficiency, providing a good way of sending just asmuch power as is required by the implant. This practice limitslosses when using a linear ohmic regulator or when using a buckconverter at low power [9].

It is tempting at first to try to maximize (6) by maximizingthe two parts of its denominator, but further analysis shows thatthe maximums of these functions do not match the maximumof the whole efficiency function. It is, therefore, more accu-rate and easier to find the global maximum of (6) numericallywith respect to the parameters of the inductive link. In partic-ular, capacitor has a strong impact on efficiency and must betuned up precisely. Fig. 4 shows the theoretical maximum of ef-ficiency for a range of practical loads and capacitor for a setof link parameters presented in Section V. It is seen that there isa maximum in and . Another very important parameternot shown is the coupling coefficient which will vary according

Fig. 4. Efficiency of the inductive link as a function of load resistance andsecondary’s capacitor.

to change in distance and lateral displacements. The efficiencywill increase monotonically with the coupling coefficient [10].

III. MULTIPLE CARRIERS TOPOLOGY

Federal Communications Commission (FCC) regulationsallow higher power densities at frequencies less than 1 MHzbecause of the lower specific absorption rate of the tissues. Fora fixed frequency, one must choose the largest possible induc-tance. Equivalently, for a fixed inductance, choosing the highestpossible frequency will maximize efficiency. The frequenciesof the data links are set to 13.56 MHz, an industrial-scien-tific-medical (ISM) band, and just as for power transmission,they rely on the use of a near-field magnetic coupling.

Crosstalk is the unwanted interference of closely pathed sig-nals. A possibility to minimize this phenomenon would be touse available technologies from multiuser channels, such as or-thogonal codes or very good filters. Since there is power con-sumption and limited processing power, however, these solu-tions are discouraged. Instead, an effective geometric approachcould provide an easy, yet very effective solution. We presentour numerical toolchain used to produce the results of this paperand then compare the orthogonal geometry [1] with our simplebut convincing coplanar arrangement.

A. Inductance Numerical Calculations

The presented coil geometries were generated with a Pythonscript we developed expressly for this purpose. The script ac-cepts parameters to allow sweeping over parameters, such as thenumber of turns, turn spacing, conductor width, and the relativeplacement of the coils. The geometry is then fed to Fasthenry-2,a quasistatic 3-D magnetic-field solver, from which inductanceand mutual inductance values can be extracted. An initial testgeometry has been validated with Agilent ADS and the resultswere found to match.

The toolchain is completed by interfacing the geometry-gen-erating Python script and Fasthenry-2 with a MATLABfunction. This function can then be plotted or optimized usingMATLAB’s pattern search tool from the direct search toolbox.


Fig. 5. Orthogonal coil geometry.

Fig. 6. Crosstalk for orthogonal coil geometry.

The optimization can then be made with respect to gain, effi-ciency, bandwidth, or any calculable variable from inductancevalues found by numerical integration. Silay presents a method-ology to improve the efficiency of an inductive link with thisprocess [11].

B. Orthogonal Coils Geometry

The three dimensionality of space allows for three mutuallyorthogonal geometries concentrated at the same point. The coilsof Fig. 5 are in this possible configuration. We can see from theillustration how symmetry is responsible for removing crosstalkfrom power into data and between the two data paths. Fig. 6shows how interference from power into data comparesto the wanted coupling between two data coils . Afteronly 3 mm of lateral displacement, both contributions are equal

, that is, the ratio is 1. This small dis-placement is easily obtained by softly moving the scalp withone’s fingers and, as such, is unacceptable for this biomedicalapplication. Furthermore, these coils are harder to fabricate thansimple planar coils.

C. Coplanar Coils Geometry

We propose using a simpler but more displacement tolerantgeometry illustrated in Fig. 7. The characteristics of these coilsare shown in Table I. We made use of coplanar square coils asthey are well documented and easier to fabricate on a numberof substrates [12]. and represent the outer and innersize of the square coils. Parameters , , , and

represent the width of the track, the spacing betweenthem, their thickness, and the number of turns on the coils,respectively. Fig. 8 shows that a spatial separation of coils is

Fig. 7. Proposed coplanar coil geometry.

Fig. 8. Crosstalk for coplanar coil geometry.

better at keeping the power coils from interfering into the datacoils under misalignment than an orthogonal approach.Data coupling is seen to be Gaussian shaped and the ratio

is less than 15% for 5-mm displacements. This resultis much better than for the orthogonal geometry; however, itshould be mentioned that the surface footprint is also increasedand that may render a multicoil coplanar system unusable oncertain areas of the body.

IV. HIGH-SPEED OQPSK DATA TRANSMISSION

Our data-transmission system presented in [5] is based on anOQPSK modulation which requires that a sine-wave carrier at13.56 MHz (ISM band) be sent through an inductive link. Here,we present a design methodology for choosing the componentvalues of the said inductive link to simplify the mathematicalanalysis and predict gain and the maximum-achievable data rate.

Before deriving any equation for gain and data transfer, letus recall that PSK modulation requires the channel, in our casean inductive link, to quickly propagate phase shifts. This delayplus the time the receiver’s phase-locked loop (PLL) needs toproperly lock onto the new phase will determine the maximaltheoretical data rate of the system. This observation allows usto immediately make critical design choices.

Fig. 9 shows the retained inductive-link topology for datatransmission. It is equivalent to the one shown in Fig. 2 exceptfor the differential structure and the additional load coupling ca-pacitor . This is due to the fact that our fabricated OQPSKchip has a differential input and output and because we needto center the signal around 900 mV at the receiver end. This isachieved by feeding both output paths to resistive divisors, notshown on the schematic for simplicity and represented by theload resistance , which has a very high value (around k )


TABLE ITHEORETICAL CHARACTERISTICS OF PROPOSED COPLANAR COIL GEOMETRY.

Fig. 9. Schematic of the inductive link used for data transmission.

in comparison with the load of the power transfer inductive link.This is required so that the receiver does not load the tank. Inequation form, we have

(10)

On the primary side, capacitor is not used to make thetank oscillate as this would impede phase changes forced bythe transmitter. It is rather used as a dc blocking capacitor tomake sure that only the impedance of inductor is seen at13.56 MHz, ensuring that the low resistance cannot short thedifferential outputs from to ground. As such, is chosenso that

(11)

In addition, the parasitic resistor of the receiver coilis increased so that oscillations before a phase shift can die outquickly, letting the new phase take over faster. This will, how-ever, put a restriction on gain and dissipate more energy, as willbe shown shortly. By design, the inductors and are ap-proximately equal and will both be denoted as simply . Sim-ilarly, because and can be dropped, will be renamed

, and will be denoted . Taking all of these design rulesand simplifications together, it is possible to rewrite (1) as

(12)

This approximation is valid for data transmission, but cannotbe applied to power-transfer analysis. Fig. 10 shows the ratio ofmagnitudes of the theoretical gain to the approximated gain as afunction of frequency and damping coefficient around the regionof interest. It shows that an error of less than 7% is introduced onthe limits of the approximation. The error is around 3% wherewe actually use this approximation.

Equation (12) is very similar to a single-tuned bandpass filtertransfer function, but for gain and an additional integration termdue to the inductive coupling. Nakajima has derived an analyt-ical solution to a carrier experiencing a discrete phase shift of

Fig. 10. Ratio of the magnitudes of theoretical gain to the approximated gain.

when passed through a bandpass filter in [13], whose fre-quency response is given by

(13)

(14)

The notation is adapted for the relevant to this present workand taken for the special case that the carrier frequency is equalto the natural frequency of the receiver’s LC-tank. Note that thisfrequency is itself dependent on the coupling coefficient asexpressed in (15), but since data coils are small with respectto their separation distance, yielding a low coupling coefficient

, the two frequencies are very close together whencompared to the bandwidth of the signal and will be approxi-mated as equal

(15)

(16)

Finding the proper response for an inductive link isdone by adapting (14) to this work with these steps:

(17)


Fig. 11. Numerically filtered 13.56–MHz OQPSK carrier modulated at 1.25Mb/s.

Although this is an approximation, simulations of (17) inSimulink are found to be in close agreement with that of (1),showing that (14) is less appropriate for an inductive link. SinceOQPSK modulation phase shifts are limited to , it makessense to calculate a time delay of a 90 phase jump definedby the time it takes for the signal in the receiver’s LC-tank tochange to 45 . From this point, a noiseless hard decision onphase could be made and corrected. Letting 0,and recalling that a 45 phase shift implies that the real andimaginary components of (17) have the same magnitude, we canderive a formula for this time delay

(18)

(19)

where is the damping coefficient of the serial RLC circuit ofthe secondary as defined by (19) and . In simula-tion or experimentally, it is hard to tell the moment at which thephase has progressed halfway to its final value. It can be shown[13], however, that this moment also corresponds to a minimumin envelope amplitude of the signal, giving a good way of evalu-ating this time delay for large . Fig. 11 shows a numericallyfiltered OQPSK carrier modulated at 1.25 Mb/s, measured atthe output of the real inductive link to illustrate this transitionalminimum envelope amplitude. The unfiltered version had irrel-evant noise in the hundreds of megahertz range.

Equation (18) shows the very important result that with a unitdamping coefficient , the time delay is for a phase shiftthat goes down to zero. This result is due to the fact that theprevious forced oscillation stops and dies exponentially withoutoscillating, that is, it is critically damped. The phase of the newforced oscillation can then takeover instantaneously. Equation(18) is not valid for as a quick verification will tell be-cause no oscillations of the previous phase are present in theLC tank during its decay. We also define the normalized timedelay with respect to the period in (20), which isplotted in Fig. 12 and compared with the normalized time delayfor a single-tuned bandpass filter. This function represents the

Fig. 12. Comparison of the number of cycles required for half a phase changethrough an inductive link and through a bandpass filter.

number of cycles required for a phase shift to reach halfway itsfinal value through an inductive link with damping coefficient

(20)

Steady-state gain may be inferred from (17) in this way

(21)

From (21) and (18), it is seen that the gain bandwidth productis given by the following function:

(22)

Equation (22) shows that an increase in coupling coefficientor carrier frequency (along with a readjustment of to satisfy(15)) will increase the gain bandwith (GBW) product. Further-more, the damping coefficient is seen to have an impact on thegain bandwidth product, a quality not seen in a simple band-pass filter where the GBW product is a constant with respect to

(from [13], ). Although (22) grows to infinity asgets close to 1, the gain decreases to zero and physical imple-

mentation of the PLL in the receiver will not be able to detectincreasingly fast phase changes.

V. EXPERIMENTAL RESULTS

In this section, we first report the characteristics of the fabri-cated inductive links presented in Section III. We then presentresults obtained for the efficiency of the retained coplanar ge-ometry inductive link and compare it to theoretical expectationsfrom Section II-C. Finally, we present strongly improved resultsof our OQPSK communication system based on the guidanceof the theory presented in Section IV. Our setup is shown inFig. 13.

A. Inductive Link

Both discussed coils geometries have been fabricated andmounted on plexiglass to be characterized with Agilent’s4294A precision impedance analyzer. Tables III and IV reportthe values of the coils, showing clearly that the self-resonanceof all coils is far above their operation frequency (1 MHz for


Fig. 13. Experimental setup featuring coplanar coils wired for data transmis-sion (bottom left coil pair).

TABLE IITHEORETICAL VERSUS MEASURED COUPLING COEFFICIENTS

power and 13.56 MHz for data). It can be seen that the predictedinductance value (see Table I) matches for most fabricated coilswithin a fairly acceptable margin. Their resistance, however,is somewhat higher than the expected value. This is due to thefact that we have overetched the copper during the fabricationprocess to reach the level of detail needed. The overetchingis especially remarkable on the data coils of the coplanargeometry, since their track width is thinner.

1) Comparison of Theory and Measurements: Table IIpresents the measured coupling coefficients of the two coilsgeometry and compares them with their theoretical predictions.In particular, we are mostly interested in coupling power intodata and data into data, as the ratio of these matter most tothe receivers. Two data sets have been selected as the mostmeaningful to this analysis, the first one being a near perfectalignment 0 mm) and the second one is chosen sothat theory predicts that power into data and data into datacoefficients become equal for the orthogonal geometry3.5 mm). Figs. 6 and 8 illustrate this theoretical prediction.The results of Table II clearly show that theoretical valuesand measurement for the orthogonal geometry are very close.The coplanar geometry has a little bit more discrepancy butthis can be explained by the difficulty of precisely measuringthe displacement and distance of the coils. Data coils of thecoplanar geometry are small in comparison to other coils, theyare more sensitive to distance and displacement measurementerrors.

TABLE IIIFABRICATED COPLANAR COILS PARAMETERS

TABLE IVFABRICATED ORTHOGONAL COILS PARAMETERS

2) Measurement of the Coupling Coefficient: The couplingcoefficients have been measured with a simple method from[14]. Equation (23) gives the relation between the couplingcoefficient, the inductances of the two coils considered, andthe voltage applied to the primary and measured at the opensecondary

(23)

B. Efficiency of the Inductive Link

Fig. 14 presents a comparison of theoretical efficiency versusmeasured efficiency. Theoretical predictions have been calcu-lated from (6) and simulated with Cadence’s Spectre softwarewith all known parasitics of the coils found in Table V. Thesewere found to match within 0.5%. The measured efficiency ofthe link, however, is lower than the expected value by about9.5%. This may be due to a combination of nonmodelled para-sitics and systematic measurement imprecision.

In measuring this efficiency, we have added a known resistorin series with the primary side circuit and used a differentialmeasurement to infer current flowing into the equivalent two-ports network. Instantaneous power was then calculated and av-eraged so that the real power dissipated by the network wouldbe known. Power at the load was simply taken as the squaredvoltage amplitude over the known load. Efficiency is then in-ferred by taking the ratio of these quantities. Equation (5) ex-presses the method used. For each load resistance used, we havemade eight measurements from 1.5 mW to 12 mW and averagedthe efficiency. This efficiency was fairly constant 1%) on this


Fig. 14. Theoretical efficiency versus measurements.

TABLE VINDUCTIVE LINK PARAMETERS FOR POWER TRANSFER

range of power for all loads. Finally, we have removed the re-sistor in series with the primary side and set the load to .For a peak power of 10.2 mW in the load, the input peak voltagewas 1.22 V, while for a peak power of 103 mW, the input peakvoltage was 3.84 V.

An efficiency of 61% is still very good for a coupling co-efficient of only 0.37. Cadence simulations show that it wouldbe possible to dramatically increase the theoretical efficiencyfrom 71% to 87% if care was taken to reduce the equivalent se-rial resistance of the capacitors and the coils’ resistances by afactor of 2. This could be possible by using a thicker copper de-posit ( oz/ft was used) and a specialized process that couldachieve the same level of detail.

C. OQPSK Data Transmission

Without the analysis presented in Section IV, we haveachieved a 2.5-Mb/s data rate with our OQPSK modulesthrough the inductive link. This data rate is in league with otherimplantable systems operating within this frequency band butcan be drastically improved through a proper analysis and tuneup of the inductive link. In Section IV, we have presented amathematical development that leads to theoretical predictionsconcerning the gain and minimum time delay for a phase-shiftpropagation through an inductive link. In simulation, (17) and(18) have been validated.

The main difference in the lab is the presence of known andunknown parasitics that makes it harder to use Fig. 9 directly.As an example, we use almost a foot length of coaxial cable tocarry the signal of the secondary LC tank to the OQPSK moduledifferentially. The capacitance of one of these cables is around35 pF and their inductance is around 400 nH. These parasiticsshift resonance and form higher order filters. Furthermore, whenthe coupling coefficient is adjusted above , the resonancefrequency of the LC tank begins to shift down progressivelymore significantly. Since (17) and (18) are good approximationswhen the carrier frequency is close to the resonance frequency,

TABLE VIINDUCTIVE LINK PARAMETERS FOR DATA TRANSMISSION

Fig. 15. Experimental validation of a 4.16-Mb/s data rate through our inductivelink.

all of these facts reduce the precision of these approximations.One can still, however, deduce good design guidance from theseequations. This has allowed us to increase the data rate up to4.16 Mb/s while, at the same time, achieving a good BER withbiomedical implants standards of .

Experimentally, a peak-to-peak signal amplitude of around600 mV has been found to be satisfactory for the PLL to per-form well at high speeds. Thus, the first step in optimizing oursystem was to adjust capacitor to reach a resonance aroundthe carrier frequency and then adjusting resistor to obtaina signal amplitude of about 600 mV. To further optimize thecommunication at high speeds, this process has to be repeatediteratively while monitoring the BER for a minimum. After onlythree iterations, a setting was found with components values il-lustrated in Table VI, allowing a communication rate of 4.16Mb/s (see Fig. 15). Further efforts include reducing the distancebetween the inductive link and the OQPSK modules, as wouldbe the case in a real implantable device, thus greatly reducingthe unknown parasitics and possible increasing data rate.

VI. CONCLUSION

We have presented experimental results matching simulationsfor our novel coil topology using a multiple carrier inductivelink. We have shown that the coplanar geometry provides betterimmunity to crosstalk under lateral misalignment. Implemen-tation of the power transfer was achieved up to an efficiencyof 61%. We have developed an analytical formula for a signalpassing through an inductive link and experiencing a discretephase shift. Finally, we have deduced from it an experimentalmethod that was used to push our OQPSK devices’ data rate to4.16 Mb/s while reducing the BER to less than . Ourresearch efforts will now focus on reducing the parasitics of the


experimental setup and including a rectifier in the power transferchain to power the OQPSK module from it.

ACKNOWLEDGMENT

The authors would like to thank CMC Microsystems for thedesign tools.

REFERENCES

[1] M. Ghovanloo and S. Atluri, “A wide-band power-efficient inductivewireless link for implantable microelectronic devices using multiplecarriers,” IEEE Trans. Circuits Syst. I: Reg. Papers, vol. 54, no. 10, pp.2211–2221, Oct. 2007.

[2] W. Liu, M. Sivaprakasam, G. Wang, M. Zhou, J. Granacki, J. Lacoss,and J. Wills, “Implantable biomimetic microelectronic systems de-sign,” IEEE Eng. Med. Biol. Mag., vol. 24, no. 5, pp. 66–74, Sep./Oct.2005.

[3] L. Jung, P. Byrnes-Preston, R. Hessler, T. Lehmann, G. Suaning, andN. Lovell, “A dual band wireless power and FSK data telemetry forbiomedical implants,” in Proc. IEEE 29th Annu. Int. Conf. Engineeringin Med. Biol. Soc., Aug. 2007, pp. 6596–6599.

[4] M. Sawan, Y. Hu, and J. Coulombe, “Wireless smart implants dedi-cated to multichannel monitoring and microstimulation,” IEEE CircuitsSyst. Mag., vol. 5, no. 1, pp. 21–39, 2005.

[5] Z. Lu and M. Sawan, “An 8 mbps data rate transmission by inductivelink dedicated to implantable devices,” in Proc. IEEE Int. Symp. Cir-cuits and Systems, May 2008, pp. 3057–3060.

[6] G. Simard, M. Sawan, and D. Massicotte, “Novel coils topology in-tended for biomedical implants with multiple carrier inductive link,” inProc. IEEE Int. Symp. Circuits and Systems, May 2009, pp. 537–540.

[7] R. O’Handley, J. Huang, D. Bono, and J. Simon, “Improved wireless,transcutaneous power transmission for in vivo applications,” IEEE Sen-sors J., vol. 8, no. 1, pp. 57–62, Jan. 2008.

[8] R. Harrison, “Designing efficient inductive power links for implantabledevices,” in , Proc. IEEE Int. Symp. Circuits and Systems, May 2007,pp. 2080–2083.

[9] J. Lee, G. Hatcher, L. Vandenberghe, and C.-K. K. Yang, “Evaluationof fully-integrated switching regulators for CMOS process technolo-gies,” IEEE Trans. Very Large Scale Integr. Syst., vol. 15, no. 9, pp.1017–1027, Sep. 2007.

[10] C. Zierhofer and E. Hochmair, “Geometric approach for coupling en-hancement of magnetically coupled coils,” IEEE Trans. Biomed. Eng.,vol. 43, no. 7, pp. 708–714, Jul. 1996.

[11] K. Silay, C. Dehollain, and M. Declercq, “Improvement of power effi-ciency of inductive links for implantable devices,” Research in Micro-electronics and Electronics, 2008. PRIME 2008. Ph.D., pp. 229–232,Apr. 2008.

[12] U.-M. Jow and M. Ghovanloo, “Design and optimization of printedspiral coils for efficient inductive power transmission,” in , Proc. 14thIEEE Int. Conf. Electronics, Circuits and Systems, Dec. 2007, pp.70–73.

[13] M. Nakajima, “Generalisation of the bandpass-lowpass analogy and itsapplication to a phase-shift-keying signal,” Electron. Lett., vol. 8, no.22, pp. 547–548, 1972.

[14] A. Salim, A. Baldi, and B. Ziaie, “Inductive link modeling and de-sign guidelines for optimum power transfer in implantable wirelessmicrosystems,” in Proc. IEEE 25th Annu. Int. Conf. Engineering Med.Biology Soc., Sep. 2003, vol. 4, pp. 3368–3371.

Guillaume Simard (S’04) received the B.Sc. de-gree in electrical engineering from the Universityof Quebec at Trois-Rivières, Trois-Rivières, QC,Canada, in 2008 and is currently pursuing theM.Sc. degree in microelectronics from PolystimNeurotechnologies Laboratory, Ecole Polytechniquede Montréal, Montréal.

His research interests include efficient power de-livery to biomedical implants and low-power high-speed telemetry.

Mohamad Sawan (S’88–M’89–SM’96–F’04)received the Ph.D. degree in electrical engineeringfrom Universite de Sherbrooke, Sherbrooke, QC,Canada, in 1990.

He joined Ecole Polytechnique de Montréal,Montréal, in 1991, where he is currently a Professorin microelectronics and biomedical engineering.His scientific interests are the design and test ofmixed-signal circuits and systems, digital andanalog signal processing, modeling, integration, andassembly. He holds the Canada Research Chair in

Smart Medical Devices, and he is leading the Microsystems Strategic Allianceof Quebec (ReSMiQ). He is Editor of the Springer Mixed-Signal Letters andAssociate Editor of the IEEE TRANSACTIONS ON BIOMEDICAL CIRCUITS AND

SYSTEMS.Dr. Sawan received the Barbara Turnbull 2003 Award for spinal cord research,

the Medal of Merit from the President of Lebanon, the Bombardier Medal ofMerit from the French Canadian Association for the Advancement of Sciences,and the American University of Science and Technology Medal of Merit. Dr.Sawan is Fellow of the Canadian Academy of Engineering and the EngineeringInstitutes of Canada. He is also “Officer” of the National Order of Quebec.

Daniel Massicotte (S’91–M’94–SM’08) receivedthe B.Sc.A. and M.Sc.A. degrees in electrical engi-neering and industrial electronics from the Universitedu Quebec a Trois-Rivières (UQTR), Trois-Rivières,QC, Canada, in 1987 and 1990, respectively, andthe Ph.D. degree in electrical engineering from theEcole Polytechnique de Montréal, Montréal, QC,Canada, in 1995.

In 1994, he joined the Department of Electricaland Computer Engineering, UQTR, where he is cur-rently a Professor. He is also Head of the Laboratory

of Signal and Systems Integration at UQTR and Chief Technology Officer ofAxiocom, Inc. He is the author or coauthor of more than 95 technical papersin internationals conferences and journals as well as author or coauthor of fivepatents pending.

Dr. Massicotte received the Douglas R. Colton Medal for Research Excel-lence awarded by the Canadian Microelectronics Corporation, the PMC-SierraHigh Speed Networking and Communication Award, and second place at theYear 2000 Complex Multimedia/Telecom IP Design Contest from Europracticein 1997, 1999, and 2000, respectively. He is also a member of the “Ordre desIngenieurs du Quebec,” “Groupe de Recherche en Electronique Industrielle,”(GREI) and “Microsystems Strategic Alliance of Quebec” (ReSMiQ).

High-speed OQPSK and efficient power transfer through inductive link for biomedical implants

Documents