See discussions, stats, and author profiles for this publication at: https://www.researchgate.

net/publication/230741633

A Power and Data Link for a Wireless-Implanted Neural Recording System

Article  in  IEEE transactions on bio-medical engineering · August 2012

DOI: 10.1109/TBME.2012.2214385 · Source: PubMed

CITATIONS READS
78 637

2 authors, including:

Alexander Rush
Illinois Institute of Technology
8 PUBLICATIONS   131 CITATIONS   

SEE PROFILE

All content following this page was uploaded by Alexander Rush on 02 April 2015.

The user has requested enhancement of the downloaded file.

 A Power and Data Link for a Wireless
Implanted Neural Recording System
Alexander Rush, Student Member, IEEE, EMBS, Philip R. Troyk, Senior Member, IEEE, EMBS

Abstract—A wireless cortical neural recording system with
a miniature implanted package is needed in a variety of
neuroscience and biomedical applications. Toward that end we
have developed a transcutaneous two-way communication and
power system for wireless neural recording. Wireless powering
and forward data transmission (into the body) at 1.25Mbps is
achieved using an FSK modulated Class E converter. The
reverse telemetry (out of the body) carrier frequency is
generated using an Integer-N PLL providing the necessary
wide-band data link to support simultaneous reverse telemetry
from multiple implanted devices on separate channels. Each
channel is designed to support reverse telemetry with a data
rate in excess of 3Mbps, which is sufficient for our goal of
streaming 16 channels of raw neural data. We plan to
incorporate this implantable power and telemetry system in a
1cm diameter single-site cortical neural recording implant.
Sending raw neural data from the implant allows one to
flexibly change the spike sorting/spike detection algorithms
in extracorporeal-based software, but comes at the expense
of high data rate requirements. To send raw neural data from
16 channels, assuming an ADC resolution of 8bits/sample
and a sampling rate of 20kSamples/s, a reverse telemetry
data rate of at least 2.56Mbps would be required.
To avoid the bandwidth limitations of an LSK system, it
is necessary to have another link for reverse telemetry. This
second link can use radiated emissions, optical coupling, or
inductive coupling to send data out of the body. For our
design, we chose to use an inductive link for reverse
telemetry.

I. INTRODUCTION
any neuroscience researchers as well as emerging
M prosthesis designs remain limited by the unavailability
of wideband transcutaneous wireless neural recording.
In the case of neuroscience research on animal subjects, an
advantage of wireless neural recording is to remove the
effect of tethering on the animal’s behavior. For neural
prosthesis design, wireless recording has the advantage of
reducing the risk of infection as well as device breakage.
A wireless neural recording system requires power and
forward data to be transferred to the implant, and neural
recording data to be transmitted from the implant. Both
power and forward data can be implemented on a single
inductive link, with forward data encoded as modulation of
the power carrier. Reverse telemetry can be achieved on the
same link by a method called load shift keying (LSK), but
the data rate is generally limited to a fraction of the power
carrier frequency, which is typically in the low MHz range. Figure 1. (a) Schematic of bidirectional data transfer system (b) Physical
Furthermore, if multiple implants are powered by the diagram of dual inductive link coils modified from [1]
external power coil, an external controller must time-
division multiplex (TDM) the reverse telemetry from the Ideally the implant would have a single-site, or “button”
implants. geometry [2], which would simplify implantation, and
prevent potential complications resulting from tethering
between multiple sections of the implant. For all wireless
neural recording implants with multiple sections of which
the authors are aware, one section resides on the skull or in a
Manuscript received February 15, 2012. This work was supported by
private donations to the Illinois Institute of Technology. bone-seat and is tethered to the microelectrode array inserted
A. Rush is with the Illinois institute of Technology, Chicago, IL 60616 in the cortex. An increased foreign body response has been
USA (e-mail: rushale@iit.edu). observed in brain tissue to implants tethered to the skull
P. R. Troyk, is with Illinois institute of Technology, Chicago, IL 60616
USA (e-mail: troyk@iit.edu).
[3,4]. Another possible complication is that the tethering
connection between multiple sections of the implant can fail
due to wire breakage or deinsulation [5,6].
To achieve a single-site geometry for a dual inductive
link, the power and data link must operate in the same
volume-space. This necessitates the consideration of
magnetic interactions between the power and data coils,
because destructive paths of the reverse telemetry signal (out
of phase with the constructive paths) can greatly reduce the
amplitude of the signal received by the external data coil.
This approach was reported by [7] for the design of a dual
inductive link for power and forward data transmission for a
retinal prosthesis.
Design of both the power and data inductive links can be
facilitated with the help of an analytic model of the inductive
link electrical and performance parameters in terms of the
link physical parameters [8-10]. This allows the physical
parameters to be iterated on a computer rather than on the
bench to find the optimal design within the physical
restriction imposed. An analytic model of the link was used
here to find the data coil radius which maximizes the
effective coupling coefficient between the data coils, taking
into account the contributions of the constructive and
destructive reverse telemetry coupling paths between the
data coils.
It is highly beneficial to lock the reverse telemetry carrier
to a multiple of the power carrier frequency using a phase-
locked loop (PLL). This provides a convenient method for
supporting simultaneous reverse telemetry from multiple
implants powered by the same magnetic field. One can
simply assign a different frequency division ratio to each
implant. This method can also simplify demodulation of the
reverse telemetry, because one can derive the reverse
telemetry carrier from the power carrier frequency.
Therefore, we have incorporated an Integer-N PLL into our
integrated circuit design, which can generate outputs of 50,
60, 70, 80, 90, and 100MHz, from the 5MHz power carrier.
The PLL cell design consumes less than 1.3mW below
100MHz, uses self-biasing techniques for supply rejection,
and has dimensions of 350um x 680um.
Fig. 1 shows a schematic of our power and bidirectional
data transfer system. In this paper we present, our dual
inductive link design methodology, implant and external
circuitry design, as well as simulation and measurement
results. Portions of this work have been previously
presented in conference form [1,11,12].
II. SYSTEM IMPLEMENTATION
The physical arrangement of the coils is illustrated in Fig.
1(b). For a typical implanted device, Coil 1 (L1) would be
the external power coil, Coil 2 (L2) would be the implanted
power coil, Coil 3 (L3) would be one of the external
differential data coils, and Coil 4 (L4) would be the
implanted data coil and is concentric to Coil 2. Power
transfer to the implant is achieved by generating a large AC
current in Coil 1 using a Class E converter. AC current is
induced in Coil 2, which is proportional to the coupling
coefficient between the external and implanted power coils,
k12. The resulting AC voltage is rectified to supply the
application-specific integrated circuit (ASIC) with power
and is also used to generate a reference clock for the ASIC.
Forward data transfer is achieved by FSK modulation of the
5MHz power carrier at a data rate of 1.25Mbps in order to
send control data to the ASIC.
In the ASIC circuitry, the reference clock, derived from
the 5MHz power carrier, is multiplied up by an integer-N
PLL to generate a reverse telemetry carrier between 50MHz
and 100MHz. The reverse telemetry is either amplitude-
shift keying (ASK) or binary phase-shift keying (BPSK)
modulated. On-chip driver circuitry induces current in Coil
4 to generate the reverse telemetry signal. According to
simulation in PSpice A/D via OrCAD Capture CIS
(Cadence Design Systems, San Jose, CA), with a power
supply of 3V, the driver circuitry can drive 2.5mA peak-to-
peak current in Coil 4.
Data is received by one of the two external differential
data coils, Coil 3. A differential coil configuration is used to
cancel both the large power signal at its fundamental
frequency and harmonics generated by the Class E converter
that fall within the frequency range of the reverse telemetry.
III. ANALYTIC MODEL
In order to avoid time-consuming design iterations on the
bench, the dual coil link for power and reverse telemetry,
illustrated in Fig. 1a,b, can be optimized with an algorithm
which iterates the modifiable link parameters and chooses an
appropriate combination of physical parameters which are
associated with the best performance, as predicted by an
analytic model of the link.
This algorithm uses an expression similar to that
presented in [7] for a dual coil system to provide power and
forward telemetry to a retinal prosthesis. The expression
was adapted for a dual coil system to provide power and
reverse telemetry. The variables used for the electrical
parameters of the link are the same as illustrated in Fig. 1a.
The derivation is similar to [7], and space does not allow
it to be included here. The assumptions critical to the
derivation of the simplified equation for data magnitude, (1),
are high quality factor coils and that the power coils are
effectively short-circuited at the data carrier frequency (e.g.,
by a parallel capacitance). This expression for the
magnitude of the reverse telemetry signal received by the
external data coil, V3, shown in Fig. 1a is
V3   I 4 jdatakeff L3 L4 (1)
k eff  k34  k 24k 23  k14k13  k14k12k 23  k 24k12k13
 where I4 is the current induced in the implant data coil by chose 8mm as the optimal diameter for Coil 4 for the 10mm
the coil driver circuitry, data is the angular frequency of the diameter of Coil 2 assumed.
In order to test the analytic model, we measured V3 as a
reverse telemetry carrier, and L3, L4, k12, k13, etc. are as
function of separation, and compared the measured values to
indicated in Fig. 1a. V3 and I4 are in phasor notation, so the
the values calculated with (1) and the equations for self and
‘j’ in (1) indicates that the steady state sinusoidal voltage on
mutual inductance as a function of the link physical
Coil 3 leads the inverse of a sinusoidal current in Coil 4 by
parameters. Again, the parameters listed in Table I were
90 degrees. According to (1), the data link can be optimized
assumed, and Coil 4 was made with a diameter of 8mm. As
by maximizing the effective coupling coefficient, keff. By
shown in Fig. 3, the measured values closely match the
analyzing the dual coil link with this equation for k eff, we
calculated values. These measurements were made with a
found that the optimal ratio of implanted power and data coil
test board designed to minimize parasitics, and a custom
radii for our design was close to 0.8.
XYZ positioning system, which has been fabricated for
To raise confidence in this idealized expression for the
testing inductive link systems in our laboratory. This XYZ
effective coupling coefficient between the data coils given in
positioning system consists of three manual linear actuators
(1) as a performance metric, we compared values of V 3
fastened together. Each linear positioner has a millimeter
simulated in PSpice A/D via OrCAD Capture CIS (Cadence
scale for accurate measurement.
Design Systems, San Jose, CA) including non-ideal,
parasitic coil parameters (effective series resistance and self-
capacitance) to values of V3 calculated using the idealized
equation (1) for ten values of implanted data coil radius,
fixing all other physical parameters. The non-ideal coil
parameters used for simulation, were calculated using our
analytic model for these parameters, presented in [1], which
space does not allow to be included here. The physical
parameters which were assumed for the data presented in
Fig. 2, while the radius of the implanted data coil was
varied, are summarized in Table I. The current in the data
coil was modeled as a sinusoid with a peak amplitude of
1mA, and coil separation was set to 1cm.

Figure 3. Comparison of measured values for reverse telemetry data signal

amplitude and the values calculated using (1) and the equations for self and
mutual inductance as a function of the link physical parameters.

More details on the analytic model of the dual inductive

link, such as coil self- and mutual-inductance, self-
capacitance and effective series resistance (ESR)
calculations are given in [1].

TABLE I. PHYSICAL PARAMETERS ASSUMED FOR DESIGN

Coil 1 Coil 2 Coil 3 Coil 4
9 0.42 0.043 0.42
Length Length Length Length
mm mm mm mm
Length
5 19 Varied,
Radius 3cm Radius of Long Radius
mm mm See Fig. 2
Figure 2. Comparison of simulated values for reverse telemetry data signal Side
amplitude using non-ideal coil parameters (effective series resistance and N.A.- Length
Insulation Insulation 5 15 Insulation 5
self-capacitance) and the values calculated using the equation (1), which Litz of Short
Thickness Thickness μm mm Thickness μm
was derived assuming nearly ideal coils [1]. Wire Side
Wire Wire 25 Trace 0.51 Wire 25
N.A.
As shown in Fig. 2, the simulated values for V3 using non- Diameter Diameter μm Width mm Diameter μm
ideal coil parameters (effective series resistance and self- Turns Per Turns Per
Turns
Turns Per
capacitance) closely match the values calculated for V3 using 3.5 12 Per 1 12
Layer Layer Layer
Layer
(1), which was derived assuming nearly ideal coils (high Q, # of # of # of # of
negligible self-capacitance). Based upon these results, we 2 3 1 1
Layers Layers Layers Layers
 TABLE II. INDUCTANCE VALUES [1]
Electrical Parameter Theoretical Measured
L1 — 4.62μH
L2 31.7 μH 32.4μH
L3 0.0573μH 0.055μH
L4 2.76uH 2.98uH
IV. COIL FABRICATION AND MEASUREMENT
The data coil and power coil were wound upon a custom-
fabricated coil form using 50 American wire gauge (AWG)
gold wire and subsequently wire bonded to a printed circuit
board (PCB) for testing of electrical parameters and
interfacing with the implant circuitry. Under the assumption
that inductances and coupling coefficients are primarily
determined by coil geometry and spacing, these parameters
were measured at 1MHz with a 1260 Impedance/Gain-
Phase Analyzer (Solartron Analytical, Farnborough, UK).
The measured and theoretical values of the coil inductance
are given Table II. The inductance of the external power
coil, L1, was measured from an existing Class E inductor in
use.
V. CLASS E CONVERTER
The magnetic field for inductive powering was generated
by a Class-E converter transmitter operating at 5MHz. The
transmitter coil carried a peak current of 0.65A, had a radius
of 3cm and 8 turns of 2MHz litz wire (New England Wire
Corporation, Lisbon, NH).
Due to the large size of the power signal compared to the
reverse telemetry signal, even small amounts of harmonic
distortion, occurring at integer multiples of the power carrier
frequency, can obscure the reverse telemetry signal.
Another source of interference can be the transmitter gate
drive, which can couple to the external data coil from the
gate-drain capacitance of the Class-E field-effect transistor
(FET). Harmonic distortion resulting from normal operation
of the Class-E converter and from the gate drive signal is
illustrated in Fig. 4.
Figure 4. Class E harmonic interference during normal operation and from
coupling of the gate drive signal into the series LC branch of the Class E
converter [11].
We have explored two different approaches to reduce
harmonics in the external power coil. One method is to
place a low-pass filter in the series tank circuit as illustrated
in Fig. 5a. The other approach, illustrated in Fig. 5b, is to
place a notch filter in the series tank of the Class E converter
to attenuate the harmonic distortion at the reverse telemetry
carrier frequency. Due to the small ratio between the reverse
telemetry carrier frequency and the power carrier frequency,
the corner frequency of the low-pass filter could not be
brought low enough to attenuate the harmonic distortion
significantly without disrupting the operation of the series
resonant tank of the Class E converter. Therefore, we chose
to use the notch filter method. Using this approach, the 12 th
harmonic (60MHz) which coincides with the reverse
telemetry carrier, was attenuated by 15dB.
Figure 5. Methods of filtering Class E harmonics from the series LC
branch of the Class E tank circuit (a) Low-pass filter in the series tank
circuit (b) Notch-filter in series tank circuit [11].
VI. DIFFERENTIAL ANTENNA
The external data receiver chosen was a pair of “bucked”
coils connected in parallel and anti-phase. In other words
the inner leads were connected together and outer leads were
connected together and grounded. This has the effect of
canceling both distant sources of RF magnetic interference
as well as nulling the 5MHz power carrier provided that the
bucked coils are carefully aligned with Coil 1. The receiver
coils could have been connected in series, in what is known
as a “figure-8” configuration. However, we found that this
made our receiver front-end susceptible to noise and
feedback. Therefore, we used the parallel coil configuration.
However, this required that we place a high-pass filter in
series with each of the bucked coils to minimize induced
power-carrier current which would have loaded the
transmitter and reduced the powering magnetic field at the
implant. A photograph of the differential reverse telemetry
receiver antenna is shown in Fig. 6.
The detection of the reverse telemetry data signal is
maximal at the center of either of the bucked coils and very
small at the shared edge of the bucked coils. The
cancellation of harmonics generated by the Class E
converter, which fall within the bandwidth of the reverse
telemetry signal, is illustrated in Fig. 7.

Figure 6. Photograph of differential reverse telemetry receiver.
Figure 7. Illustration of harmonic interference nulling by the differential
reverse telemetry receiver [11].
VII. IMPLANT CIRCUITRY
An application specific integrated circuit (ASIC) was
designed to implement the circuit portion of the wireless
power and data system, which, for an implanted device,
would be located inside the body. As shown in Figs. 8 & 9
the ASIC contains a fully integrated rectifier, a PLL,
modulators (ASK and BPSK), and reverse telemetry drivers.
The external circuitry for wireless powering and two-way
communication is also presented. The integrated circuit was
fabricated in the X-FAB (Lubbock, TX) 800nm BiCMOS
process [13] (BiCMOS is a term for a semiconductor
Figure 8. Schematic of Telemetry IC Multi-Channel Wireless Neural Recording System [12]
technology that integrates bipolar junction transistors and
complementary metal-oxide-semiconductor transistors).
The PLL has a programmable output frequency to allow
multiple implanted devices to send reverse telemetry from
roughly the same physical location. Specifically the
frequency divider is designed to synthesize voltage-
controlled oscillator (VCO) outputs of 50, 60, 70, 80, 90 and
100MHz depending on the value of a 4-bit control word.
Producing these frequencies required division-by-two
followed by division by 5, 6, 7, 8, 9, or 10.
Figure 9. PLL (and ASK/BPSK Transmitter in lower right) integrated
circuit (IC) Die Photograph. The total die area was 1600umx1450um
including the pads and ring. The PLL dimensions were 350um x 680um
[12].
Self-bias techniques and a differential buffer were used to
improve the power supply rejection of the VCO as described
in [14]. A more complete description of the PLL circuitry
and its measured performance is given in [12].
The 5MHz reference clock is recovered from the magnetic
power field carrier by the on-chip FSK demodulator
circuitry which also demodulates commands sent to the RF
telemetry ASIC by the transmitter via FSK modulation of
the power carrier at a data rate of 1.25Mbps as described in
[15].
 TABLE III
WIRELESS NEURAL RECORDING SYSTEMS AND CHARACTERISTICS OF THEIR POWER AND DATA SYSTEMS

Implantability Raw Neural

Wireless Neural Recording (According to Power Data Transmission Recording or Data Data Link
System authors’ stated plans) Source Method Spike Detection Rate Energy per bit

WINeR System, Georgia Radiated, ISM band Raw Neural 58-

No Inductive 607pJ/b
Institute of Technology [16] at 915MHz Recording 709kSps

Yes, Two Island Raw Neural

Brown University System [2] Inductive Optical N/A N/A
(Two-Site) Geometry Recording

INI System, U of Utah and Yes, Button (Single- Radiated, ISM Band
Inductive Spike Detection 157kbps 3185pJ/b
Stanford [17] Site) Geometry 902-928MHz

Hermes System, U of Utah Radiated,

No Battery Raw Data 24Mbps 1250pJ/b
and Stanford [18] 3.7~4.1GHz

Radiated, Impulse
Yes, At Least Two Raw Neural
UCSC System [19] Battery radio based UWB, 90Mbps 17.78pJ/b
Site Geometry Recording
4GHz

University of Michigan, Ann Yes, Two Island Coil Antenna, 70-

Inductive Spike Detection 2Mbps N/A
Arbor System [20] Geometry 200MHz

Yes, Button Coil Antenna, 50- Raw Neural

This Work Inductive 3Mbps 1962pJ/b
Geometry 100MHz Recording

TABLE III
and data link with a distance of 20mm and a data rate of
In simulation, the PLL power consumption was below 3Mbps is given in Fig. 10.
1.3mW for frequencies below 100MHz with a 3V supply.
The power consumption of the fabricated ASIC was
measured to be 4.2mW operating at 48MHz with a 2.7V
supply, and the measured power consumption closely
matched OrCAD simulations. Simulations predict that the
PLL consumes only 14% of the total power, while the
reverse telemetry driver consumes 74% of the total power.
The reverse telemetry driver circuitry included on the chip is
not optimized for low power because it has redundant
modulation and driver circuitry that was used for evaluation
purposes.

VIII. DEMODULATION OF THE REVERSE TELEMETRY

Although synchronous demodulation might be easily
achieved, since the outward RF signal carrier is phase locked
to the RF power carrier, for initial testing the reverse
telemetry signal was asynchronously demodulated. This is
achieved using a bandpass filter centered at the reverse
telemetry carrier frequency, followed by a logarithmic
amplifier with a large dynamic range.
To facilitate digital algorithm testing, the finite impulse
response (FIR) filter and data synchronization system was
implemented on the Cyclone III DSP Development Kit
(Altera, San Jose, CA). With a coil separation of 20mm and
a data rate of 3Mbps, the bit error rate (BER) was measured
to be 2.03e-4. Measured data of the dual inductive power
Figure 10. Demonstration of ASK transmission and demodulation with a
coil separation of 20mm and a data rate of 3Mbps. The top trace is the
modulating data signal to the implant circuitry for reverse telemetry. The
second trace from the top is the log amp output. The third trace from the
top is the digitally filtered (and inverted) log amp output. The bottom trace
is the demodulated reverse telemetry data.
We believe 20mm is a separation distance that is adequate
for most cortical neural-recording implants based on a
conservative estimate of the separation imposed by anatomy
for an adult male human, taking into account the thickness of
the scalp (≈8mm maximum [21]) and skull (≈11mm [22]), as
well as the dura and subdural space [23].
 Having shown that the wireless power and telemetry
system was capable of delivering the required data rate and
coil separation, the wireless power and telemetry circuitry
was then demonstrated in the context of a prototype 4
channel wireless neural recording system. The overall
architecture of a prototype wireless neural recording array
(WNRA) circuitry is shown in Fig. 11.
Figure 11. Overall architecture of our prototype wireless neural recording
array (WNRA) circuitry. This ASIC includes neural recording amplifiers
and a specialized voltage regulator to reduce power supply noise and assure
reliable operation even with expected variations in the powering magnetic
field strength. Coil A, B are the inputs connected to the leads of the implant
power coil, and Coil C, D are the outputs connected to the leads of the
implant data coil. ModSel is used to choose between BPSK and ASK
modulation. Vdd is the shunt regulated power supply and input to a low
dropout voltage regulator, the output of which is VLV.
The purpose of this prototype WNRA circuitry was to
evaluate the wireless power and reverse telemetry link in the
context of a wireless neural recording system with an
architecture representative of systems into which the power
and data link will ultimately be incorporated. The prototype
WNRA circuitry used available amplifiers and ADC
circuitry. No attempt was made to minimize the noise on the
amplifiers, nor were comprehensive noise measurements
made.
A simplified illustration of the DSP filtering and data
synchronization system used to demodulate the reverse
telemetry from the prototype four channel wireless neural
recording circuitry is given in Fig. 12. This system was
implemented on an FPGA so changes could be made quickly
by reprogramming, which is preferable to a digital ASIC or
analog system, which would require new hardware to be
ordered. The Cyclone III DSP Development Board and
Cyclone III Data Conversion HSMC (Altera Corporation,
San Jose, CA) were used for rapid prototyping purposes.
However, ultimately only an ADC, the FPGA, and a flash
memory would be required, or the design could be converted
to an equivalent ASIC or discrete circuit design.
Figure 12. Simplified block diagram of the DSP filtering and data
synchronization system used to demodulate reverse telemetry from the
prototype four channel WNRA circuitry.
Although the prototype WNRA circuitry is still under
evaluation, a set of preliminary measurements were made.
The gain and high pass filter corner of the amplifiers were
set to their lowest possible values. The gain was measured
to be 62.3 and the bandwidth was measured to be 0.7Hz to
27kHz. The waveforms were digitized by the ADC on the
prototype WNRA circuitry at a rate of 20kSps (the potential
for aliasing should be fixed in future versions of the chip
with an adjustable low pass corner) and with a resolution of
8bits/Sample.
A 10mV, 3kHz sinusoidal input was presented to the four
electrode inputs of the prototype four channel wireless
neural recording array circuitry. Reverse telemetry data was
transmitted with a coil separation of 15mm, a carrier
frequency of 40MHz, and a data rate of 1.25Mbps. The
reverse telemetry was successfully demodulated, and the
amplitude of the wirelessly transmitted and decoded
amplifier outputs were found to be within the expected
range.
IX. DISCUSSION
We have shown how a dual inductive link for
transcutaneous wireless power and two-way communication
can be optimized using an analytic model of the inductive
link in terms of its physical parameters, avoiding time-
consuming design iterations on the bench. No publication
existed previously for the optimization of reverse telemetry
achieved using a dual inductive link with coaxial implanted
coils. The coaxial coil arrangement is attractive, because the
data coil fits within the area of the power coil, and,
therefore, does not increase the total area of the implant.
Also, in contrast to the approach where the data coils are
made to be orthogonal to the power coils the thickness of the
implant is not increased by the presence of the data coils.
Also, a greater coupling coefficient is achieved with the
coaxial dual inductive link method than with the orthogonal
dual inductive link method.
Our aim was to incorporate the power and two-way
communication system into an implant with a diameter of
1cm. A small implant diameter allows more implants in a
given area of cortex, all of which could be powered by a
single external power coil. Also, implantation may be
achieved more easily with an implant having a smaller
footprint. The 1cm diameter restriction on coil diameter
presents a challenge in delivering sufficient power to the
implant and reverse telemetry amplitude outside the body,
because, with the coil separation imposed by anatomy
(mainly skull and scalp) a small implant diameter results in
small coupling coefficients between the external and implant
coils. Therefore, the analytic model for a dual inductive link
was used to achieve a working power and reverse telemetry
system having implant coil diameters of no more than 1cm.
Techniques for reducing and mitigating harmonic
interference from the power link onto the data link were also
presented, without which, the reverse telemetry signal would
be obscured. Particularly important was the use of
differential coils for detecting the reverse telemetry signal
from the implant data coil, while rejecting harmonic
interference from the external power coil. Historically,
differential coils, have been used to reject the large power
carrier rather than to reduce harmonic interference from the
power link onto a separate data link operating in the same
space. For instance, differential coils have been used in a
dual inductive link for delivering power and forward
telemetry to an implant. In this case the purpose of the
differential coil was primarily to reduce the size of the filters
that would to reject the power carrier in the implant
circuitry, where size is strictly limited by anatomy.
An end-to-end demonstration of a prototype wireless
neural recording array (WNRA) circuitry has shown that the
wireless power and reverse telemetry link is functional in the
context of a representative wireless neural recording system.
Although the preliminary measurements of the 4-channel
prototype WNRA circuitry presented here would be
considered rough for the purpose of demonstrating a
complete neural recording system, the purpose of the end-to-
end demonstration was to demonstrate the wireless power
and telemetry link which has been a historic problem. In
continuing work the amplifier and ADC need to be
optimized for noise and accurate gain control. However,
there are numerous examples in the literature of low noise,
low power amplifiers which have been incorporated into
neural recording circuitry, so we do not anticipate any
obstacles in designing amplifiers and ADC circuitry
appropriate for a first generation wireless neural recording
system [2,16-20].
Details regarding the layout, materials, and characteristics
of the electrode array, which will be used with a first
generation recording system utilizing the wireless power and
two-way communication system presented here is also
beyond the scope of this paper. However, we anticipate that
the neural recording system will be designed to work with a
variety of electrode types.
Table III summarizes several of the most developed
wireless neural recording systems for comparison and some
essential characteristics of their power and data systems.
One strength of the wireless neural recording system
presented here is the use of a single site, “button”, geometry,
which would allow for implantation of the entire recording
system beneath the dura. A single site implant does not
require connections between multiple sections of the
implant, which simplifies implantation and eliminates any
potential damage which may result from tethering and the
relative motion of the skull and brain.
The INI system, summarized in Table III, also uses a
button geometry. However, in contrast to the INI system,
which sends spike detected data, the wireless neural
recording system presented here is intended for sending raw
neural data on all channels simultaneously, which requires a
much higher data rate. The system presented here is also
designed to allow multiple implants located in the magnetic
field of a single Class E powering coil to send reverse
telemetry simultaneously on separate channels (e.g. 50, 60,
70, …, 100 MHz).
X. CONCLUSION
The system presented here is an important step for
providing two-way communication and wireless power in a
wideband transcutaneous neural recording system. The dual
inductive link was optimized using an analytic model of the
link in terms of its physical parameters. Novel methods
were used to reduce interference between the power and
reverse telemetry link, including filtering of harmonics from
the Class-E converter tank, and use of a differential reverse
telemetry receiving coil to cancel transmitter harmonic and
far-field interference.
The wireless power and data system was demonstrated in
the laboratory by fabricating an inductive link and pairing
with an ASIC. Operation is presented for a separation of
20mm and a data rate of 3Mbps, which we believe would be
sufficient for a neural prosthesis utilizing cortical neural
recordings.
REFERENCES
[1] A.D. Rush, P.R. Troyk, “Dual Inductive Link Coil Design for a
Neural Recording System,” Proc. 33rd Annu. Int. Conf. IEEE EMBS,
2011.
[2] A. Nurmikko, J. Donoghue, L. Hochberg, W. Patterson, Y. Song, C.
Bull, D. Borton, F. Laiwalla, S. Park, Y. Ming, J. Aceros, “Listening
to Brain Microcircuits for Interfacing With External World- Progress
in Wireless Implantable Microelectronic Neuroengineering Devices,”
Proc. IEEE, vol. 98, no. 3, pp. 375-388, Mar. 2010.
[3] R. Biran, D. Martin, P. Tresco, “The Brain Tissue Response to
Implanted Silicon Microelectrode Arrays is Increased When the
Device isTethered to the Skull,” Journal of Biomedical Materials
Research, vol. 82, no. 1, pp. 169-178, Jan. 2007.
[4] Y. Kim, R. Hitchcock, M. Bridge, P. Tresco, “Chronic Response of
Adult Rat Brain Tissue to Implants anchored to the Skull,”
Biomaterials, vol. 25, no. 12, pp. 2229-2237, 2004
[5] J. Hetke, J. Lund, K. Najafi, K. Wise, D. Anderson, “Silicon Ribbon
Cables for Chronically Implantable Microelectrode Arrays,” IEEE
Transactions on Biomedical Engineering, vol. 41, no. 4, Apr. 1994.
[6] R. Rennaker, A. Ruyle, S. Street, A. Sloan, “An Economical Multi-
Channel Cortical Electrode Array for Extended Periods of Recording
During Behavior,” Journal of Neuroscience Methods, vol. 142, pp. 97-
105, 2005.
[7] G.Wang, W. Liu, M.Sivaprakasam, M. Zhou, J. Weiland, M.
Humayun, “A Dual Band Wireless Power and Data Telemetry for
 Retinal Prosthesis,” Proc.28th IEEE EMBS Annu. Int. Conf., pp. 4392- Alexander D. Rush (M’09) received the B.S.
4395, Aug. 2006. degree in electrical engineering from University of
[8] U. Jow, M. Ghovanloo, “Modeling and Optimization of Printed Spiral Illinois at Urbana-Champaign in 2006 and the
Coils in Air, Saline, and Muscle Tissue Environments,” Proc. 31st Ph.D. degree in biomedical engineering from
Annu. Int. Conf. IEEE EMBS, pp. 6387-6390, Sep. 2009. Illinois Institute of Technology in 2012.
[9] P.R. Troyk, A.D. Rush, “Inductive Link Design for Miniature He is currently developing new hardware for
Implants,”Proc. 31st Annu. Int. Conf. IEEE EMBS, pp. 204-209, Sep. neural recording and stimulation at Plexon Inc in
2009. Dallas, TX. His research interests include analog,
[10] Z. Yang, W. Liu, E. Basham, “Inductor Modeling in Wireless Links digital and mixed signal circuit design,
for Implantable Electronics,” IEEE Trans. Magnetics, vol. 43, no. 10, neuroprosthetic devices, wireless power and
pp. 3851-3860, Oct. 2007. telemetry for implantable electronics, and
[11] A.D. Rush, P.R. Troyk, “Electronic Performance of a Dual Inductive optogenetics.
Link for a Wireless Neural Recording Implant,” Proc. 33rd Annu. Int.
Conf. IEEE EMBS, 2011.
[12] A.D. Rush, P.R. Troyk, “Power and Data for a Wireless Implanted
Neural Recording System,” Proc. 5th Int. IEEE EMBS Conf. Neural
Engineering, 2011. Philip R. Troyk (M’83–SM’91) received the B.S.
[13] XFAB, (2011, Feb.) 0.8µm CMOS Process (Rev. 3.3) [Online]. degree in electrical engineering from the
Available: University of Illinois at Urbana-Champaign, in
www.xfab.com/fileadmin/http://www.xfab.com/fileadmin/X- 1974, and the M.S. and Ph.D. degrees in
FAB/Download_Center/Technology/CMOS/CX08_Data_sheet.pdf bioengineering from the University of Illinois,
[14] J. Maneatis, “Low-Jitter Process-Independent DLL and PLL Based on Chicago, in 1980 and 1983, respectively.
Self-Biased Techniques,” IEEE J. Solid-State Circuits, vol. 31, no. 11, He was on the staff of Northrop Corporation,
pp. 1723-1732, Nov. 1996. Rolling Meadows, IL, from 1973 to 1981. In 1983,
[15] P.R. Troyk, G. DeMichele, “Inductively-Coupled Power and Data he joined the faculty of the Illinois Institute of
Link for Neural Prostheses using a Class-E Oscillator and FSK Technology, where he is currently Associate Dean
Modulation,”Proc. 25th Annu. Int. Conf. IEEE EMBS, pp. 3376-3379, of the Armour College of Engineering, Associate
Sep. 2003. Professor of Biomedical Engineering, and Director
[16] S. Lee, H. Lee, M. Kiani, U. Jow, M. Ghovanloo, “An Inductively of the Laboratory of Neural Prosthetic Research. He is also president of
Powered Scalable 32-Channel Wireless Neural Recording System-on- Sigenics, Inc, a company involved with design of ASICs for medical use. At
a-Chip for Neuroscience Applications,” IEEE Trans. Biomedical IIT he leads a team for development of an intracortical visual prosthesis,
Circuits and Systems, vol. 4, no. 6, pp. 360-371, Dec. 2010. and directs IIT’s contribution towards development of the IMES for
[17] R. Harrison, R. Kier, C. Chestek, V. Gilja, P. Nuyujukian, S. Ryu, B. prosthesis control. His broader interests include development of central and
Greger, F. Solzbacher, and K. Shenoy, “Wireless Neural Recording peripheral neural prostheses, the design and packaging of electronic
With Single Low-Power Integrated Circuit,” IEEE Trans. Neural assemblies for implantation in the human body, and polymeric protection of
Systems and Rehabilitation Engineering, vol. 17, no. 4, pp. 322-328, thin film devices operating in high humidity environments.
Aug. 2009.
[18] H. Miranda, V. Gilja, C. Chestek,K. Shenoy,T. Meng, “Hermes D:A
high-rate long-range wireless transmission system forsimultaneous
multichannel neural recording applications,” IEEE Trans. Biomedical
Circuits and Systems, vol. 4, no. 3, pp. 181-191, Jun. 2010.
[19] M. Chae, Z. Yang, M. Yuce, L. Hoang, and W. Liu, “A 128-Channel
6 mW Wireless Neural Recording IC With Spike Feature Extraction
and UWB Transmitter,” IEEE Trans. Neural Systems and
Rehabilitation Engineering, vol. 17, no. 4, pp. 312-321, Aug. 2009.
[20] A. Sodagar, G. Perlin, Y. Yao, K. Najafi, K. Wise, “An Implantable
64-Channel Wireless Microsystem for Single-Unit Neural Recording,”
IEEE J. Solid-State Circuits, vol. 44, no. 9, pp. 2591-2604,Sep. 2009.
[21] C.H. Raine, C.A. Lee, D.R. Strachan, C.T. Totten, S. Khan. “Skin
Flap Thickness in Cochlear Implant Patients - a Prospective Study,”
Cochlear Implants Int.,vol. 8, no. 3, pp. 148-157, Sep. 2007.
[22] A. Adeloye, K.R. Kattan, and F.N. Silverman, “Thickness of the
Normal Skull in the American Blacks and Whites,”Am. J. Phys.
Anthrop., vol. 43, no. 1, pp. 23-29, Jul. 1975.
[23] L. Manola, B.H. Roelofsen, J. Holsheimer, E. Marini, J. Geelen,
“Modelling motor cortex stimulation for chronic pain control:
electrical potential field, activating functions and responses of simple
nerve fibre models,” Med. Biol. Eng. Comput., vol.. 43, no. 3, pp. 335-
343, Jun. 2005.

View publication stats