General rights
Copyright and moral rights for the publications made accessible in the public portal are retained by the authors and/or other copyright
owners and it is a condition of accessing publications that users recognise and abide by the legal requirements associated with these rights.
Users may download and print one copy of any publication from the public portal for the purpose of private study or research.
You may not further distribute the material or use it for any profit-making activity or commercial gain
You may freely distribute the URL identifying the publication in the public portal
If you believe that this document breaches copyright please contact us providing details, and we will remove access to the work immediately
and investigate your claim.
Downloaded from orbit.dtu.dk on: Aug 10, 2022
A distributed transducer system for functional electrical stimulation
Gudnason, Gunnar; Nielsen, Jannik Hammel; Bruun, Erik; Haugland, Morten
Published in:
Proceedings on 8th IEEE International Conference on Electronics, Circuits and Systems
Link to article, DOI:
10.1109/ICECS.2001.957763
Publication date:
2001
Document Version
Publisher's PDF, also known as Version of record
Link back to DTU Orbit
Citation (APA):
Gudnason, G., Nielsen, J. H., Bruun, E., & Haugland, M. (2001). A distributed transducer system for functional
electrical stimulation. In Proceedings on 8th IEEE International Conference on Electronics, Circuits and Systems
(Vol. 1) https://doi.org/10.1109/ICECS.2001.957763
A
distributed
transducer
system
for
functional
electrical stimulation
Gunnar Gudnason, Jannik H. Nielsen and Erik Bruun
Orsted-DTU, Technical University
of
Denmark,
DK-2800
Lyngby
gg@oersted.dtu.dk
Morten Haugland
Center
for
Sensory-Motor Interaction,
Aalborg
University,
DK-9220
Aalborg
Abstract
Implanted transducers for functional electrical stim-
ulation (FES) powered by inductive links are subject to
conflicting requirements arising from
low
link eflciency, a
low power budget and the need for protection of the weak
signals against strong RF electromagneticJields.
We propose a solution to these problems by partition-
ing the RF transceiver and sensor/actuator functions onto
separate integrated circuits.
By
amplifying measured neu-
ral signals directly at the measurements site and convert-
ing them into the digital domain before passing them to the
transceiver; the signal integrity is less likely to be affected
by the inductive link. Neural stimulators are afected to a
lesser degree, but still benejit from the partitioning.
As
a test case, we have designed a transceiver and a
sensor chip which implement this partitioning policy. The
transceiver is designed to operate in the
6.78MHz
ISM
band and consumes approximately
360pW
Both chips
were implemented in a standard
0.5
,um
CMOS
technol-
ogy, and use a
3
V
supply voltage.
1.
Introduction
The subject of this article is a solution for several
problems which plague designers of inductively powered
biomedical implants for functional electrical stimulation.
Inductive links are commonly employed to power and
control such implants since they provide a wireless con-
nection, which is desirable since percutaneous wires pro-
vide an infection path into the body. Inductive links also
eliminate the need
for
an implanted battery, which might
eventually need to be replaced. The price which must be
paid for these advantages is the relatively high power dis-
sipation in the external (non-implanted) apparatus due to
the
low
efficiency of the link, and a degradation of the sig-
nals under examination, because of the high electromag-
netic fields. While the
RF
signals normally lie far outside
the biological signal band, they can easily desensitize an
amplifier designed for a
10-2OpV
signal range.
The signal amplitude quoted above is typical for sig-
nals obtained using cuff electrodes
[
11. Stimulators con-
nected to cuff electrodes generate much larger signals,
so
they are not affected to the same extent as sensors by the
external
RF
field.
We propose therefore a physical partitioning of the sig-
nal processing functions, in order to limit the transmission
w
transmitter
\
Figure
1.
A
partitioning example for a
FES
system con-
taining a transceiver chip and sensor and actuator chips.
The last two are not necessarily connected to the same
nerve trunk.
distances for sensitive signals, and to allow optimal place-
ment of critical functions of the system. The conflicting
design criteria which can be accommodated to a large ex-
tent by partitioning are:
The
RF
part of the system should be close to the skin
surface for better power transmission, and also for
better link bandwidth. The proximity may also be
an
advantage for surgical access.
The transducers should be placed close to the active
sites, which can be relatively deep inside the body.
Long routing of transducer signals should be
avoided, as the strong electromagnetic field will
induce an
RF
interferer overlaid on the desired sig-
nal. Physical separation of the transducer from the
RF
link also reduces the problems, since the field
falls off rapidly with distance (as
1
/r3
in the far-field
limit).
The system must contain a considerable amount
of
digital logic,
for
control, timing, and buffering of
data. The logic will inevitably couple switching
noise onto the supplies and into the substrate. Using
low-noise logic solutions like current-steering logic
[3]
can eliminate the problem, but the static supply
current of CSL and other ultra-low noise logic fami-
lies makes them unsuitable as the system complexity
passes a certain point. Placing the sensors on sepa-
rate chips isolates them from the logic supply noise.
One option which was examined, and which alleviates
some of the problems above, is to place only the receiver
0-7803-7057-010
1
/$
10.00
0200
1
IEEE.
397
Authorized licensed use limited to: Danmarks Tekniske Informationscenter. Downloaded on March 23,2010 at 09:13:27 EDT from IEEE Xplore. Restrictions apply.
Sensor chip
Figure
2.
A
simplified block diagram of the system, showing the main parts of the transceiver and sensor chips.
Differential signals are shown with a single line for clarity
coil close to the skin surface, and route the antenna signal
to a single integrated transceiver/transducer. This is how-
ever not practical since biocompatible interconnect solu-
tions like Cooper cable' have a high impedance at DC and
RF.
We adopted a solution with simple digital communica-
tion protocols between the central transceiver and control
chip, and the outlying transducer chips. The protocol was
designed
so
that it could be accommodated along with the
power supply wires in a simple 4-wire flexible cable. In
this specific test case, we implemented a simple system
with one transceiver chip and one sensor chip. The sensor
chip includes an amplifier for cuff electrode signals, which
is itself the subject of another article [2], an AD converter
and bus interface logic. The transceiver includes a direct
conversion receiver (actually a homodyne), a load modu-
lation circuit for transmission of data out of the system,
control logic, supply regulators and references.
2.
System description
The following is a description of the transceiver chip
and the aspects relating to communication between it and
the other parts of the system. Since the main focus of this
article is on the way in which partitioning of functions
can solve some of the problems in an implanted system of
this type, the sensor chip will only be described inasmuch
as it relates to the top-level design of the system. The
principle of partitioning applies to any transducer type that
is relevant in an implanted system,
so
the internals of the
sensor chip will not be emphasized. Figure 2 shows atop-
level diagram of the system.
2.1.
Thelink
The use
of
inductive links for power and data transmis-
sion is described in great detail elsewhere, see for example
By separating the transceiver chip and the antenna, it is
possible to reduce the distance between the external trans-
[5,61.
mitter antenna (a tuned LC circuit) and the internal one to
15-20mm. The coupling coefficient for normal coil ge-
ometries at this distance is on the order of
0.01-0.05.
This
permits far better power transfer than if the transceiver
were placed together with the transducer, and makes load
modulation a viable method to extract information from
the system. Systems with smaller coupling coefficients
must resort to active transmission of signals to the outside
[4], which increases the power consumption.
The chosen modulation method is PAM with Manch-
ester encoding and a modulation index of approximately
0.2. This is compatible with high-efficiency class D or
E
transmitter configurations
[7,
81.
The carrier frequency is
6.78MHz
which coincides with one of the
ISM
(indus-
trial, scientific and medical) bands. The target bit rate is
5Okbitls.
2.2.
The power supply
The power is extracted from the
RF
carrier by means
of
a hll-wave bridge rectifier, which is implemented by
us-
ing diode-connected P-channel
MOS
transistors in a com-
mon N-well, as shown in figure
3.
The standard CMOS
process does not offer high-quality floating diodes,
so
an-
other solution must be chosen. The available elements
are three types of p-n junction and diode-connected
MOS
transistors, but these all suffer from high substrate currents
and/or parasitic elements which divert some of the current
from its intended path. The
PMOS
rectifier bridge is ac-
companied by parasitic vertical PNP transistors which can
divert some of the input current to
V,,
instead of to
VDD.
'Produced
by
Finetech
Medical
Ltd.
398
Figure
3.
The power conversion circuit.
Authorized licensed use limited to: Danmarks Tekniske Informationscenter. Downloaded on March 23,2010 at 09:13:27 EDT from IEEE Xplore. Restrictions apply.
100
mA
lOmA
1
mA
100pA
10pA
lOOnA
Figure
4.
The measured
shunt
regulator current
We have solved this problem by dimensioning the
PMOS
transistors
so
that they are biased in weak inversion over
the entire operating range. Since the nominal threshold
voltage is about
0.6V,
there is a range of drain-gate volt-
ages for M1 in figure
3
(corresponding to the emitter-base
potential in the parasitic
PNP)
where the
MOS
current
dominates the current through the bipolar transistor by a
large factor. The current-handling capability
of
the recti-
fiers can be increased for a given maximum parasitidmain
current ratio, by increasing the width (and area)
of
the
transistors. The main penalty is an increased input capac-
itance. Our experimental data show that the parasitic cur-
rent ratio is smaller than
0.005
for a system supply current
of
200pA
and an input capacitance of 2pF.
All
the tran-
sistors in figure
3
have
W/L
=
800/0.5
in micrometers.
The transducer chip contains an active shunt regulator
which provides protection against excessive input power.
The transmitted power
is
a very strong function of antenna
spacing,
so
an unprotected circuit can easily be burned
out.
The
shunt
circuit
is
basically
a
pass
transistor be-
tween
V,,
and
Vss,
and a feedback loop which compares
V,,
to a bandgap reference voltage.
As
the supply volt-
age approaches the trigger point (which was set equal to
the maximum supply voltage for the technology), the reg-
ulator begins shunting current from the supply, and the
current changes by
4
decades over a short supply voltage
interval (see figure
4).
The performance of the power supply conditioning sys-
tem is shown in figure
5.
The conversion
of
the
RF
power
to a DC supply voltage and the overvoltage protection is
carried out mostly on-chip, with the only external compo-
nent being the energy storage capacitor
C,.
The transmit-
ter used in these measurements was running at a relatively
high power level, witnessed by the fact that the full supply
voltage is reached at a separation of 120mm. Despite the
strong dependence of the magnetic field amplitude on the
separation
d,
the on-chip shunt succeeds in maintaining
the supply voltage at the nominal level at all separations.
The transceiver chip provides the supply voltage for the
other chips in the system. Two strands out of four in the
Cooper cable have been assigned to
Vss
and
VD,.
Because
of the helically wound construction of the cable, and the
Pt-Ir composition, the impedance
is
relatively high at all
frequencies. To provide a stable supply voltage for the
Figure
5.
The measured on-chip supply voltage
as
a
function of the separation between the transmitter and
transceiver antennas.
outlying chips, it is therefore necessary
to
add decoupling
capacitors to the supply lines at the transducer ends.
A
stimulator chip can require relatively large current pulses
from the supply while delivering a stimulus.
2.3.
Transceiver
The receiver part of the transceiver chip is a direct-
conversion receiver. Since the input signal has a larger
amplitude than the supply range, an attenuator with a gain
of approximately
0.1
is inserted in the signal path. The
input signal is tracked by a
PLL
which also provides the
system clock (the block diagram does not show a clock
divider circuit which divides the clock frequency down to
1.7MHz).
The output of the
PLL
and the input signal are put
through a mixer, and the mixer output is filtered in a 4th-
order differential G,,< filter. The filter output is directed
to
a mixed analog-digital level detector, which detects
the sign of the transmitted data bits. Instead of using a
partly analog level detector, the normal processing method
would be to sample the filter output for further digital
sig-
nal processing. This does however require more sophisti-
cated digital circuits than we were willing to implement,
and possibly requires more supply current.
The target bit rate for the system was 50kbiv's, and the
receiver is designed to handle up to 10Okbit/s.
Data
is
transmitted out of the system by load mod-
ulation. The reflected impedance seen by the external
transmitter is varied by changing the load seen by the
secondary
LC
circuit.
A
switch is connected between
the terminals
of
the antenna, and by closing the switch,
a maximum change in the reflected impedance is ob-
tained. This simple scheme has the disadvantage of stop-
ping power transfer to the system during load modulation,
and other load modulation circuits have been designed to
avoid this
[9].
The duty cycle of the switch closure is how-
ever
so
low in our case that the reduction in average power
transfer is small.
2.4.
Interchip communication
The cable type that we used as a reference for
our
inter-
connects is, as mentioned before, a 4-strand biocompati-
399
Authorized licensed use limited to: Danmarks Tekniske Informationscenter. Downloaded on March 23,2010 at 09:13:27 EDT from IEEE Xplore. Restrictions apply.
ble cable type by the name of Cooper cable. In addition
to being biocompatible, the cable is wound in a helical
pattern and embedded in silicone,
so
it can be stretched.
The stretching reduces the probability of damage to the
surrounding tissue.
Because
of
the biological constraints that the cable
must fulfill, it does not have very good electrical prop-
erties, and the use of the cable must be adjusted accord-
ingly. We measured the series resistance at
DC
of a repre-
sentative length
of
the cable, and found it to be 2
10
Rlm.
The capacitance between any two wires varies from
20-
80pF/m because of the lack of symmetry, and the capaci-
tance
of
each wire to the surroundings is about SOpFlm.
No
clock signal is sent across the connection between
the chips, Instead, we use an asynchronous handshake
mechanism to control data transfer, and an internal oscil-
lator in the sensor chip to provide a time base.
The line drivers are class
AB
circuits which have a qui-
escent current consumption of
5pA
each, and can slew
the line voltages with a
50
pA
current. This type of driver
was chosen instead of faster types because this is more
than sufficient for the purpose, and by limiting the slew-
ing currents the supply transients are reduced. The high
and low voltages on the signal lines are
1.OV
and
0.3V
respectively instead of the full supply range, again to re-
duce supply transients and power consumption.
3.
Measurement results
Most of the basic functions of the transceiver and sen-
sor
chips have been measured, and they behave according
to specifications.
The results for the power conversion circuit show that
the use of PMOS transistors in weak inversion as rectifiers
is an ideal solution for pure CMOS technologies. Cur-
rent throught the substrate are eliminated by placing ev-
erything in an N-well, and the effect
of
parasitic bipolars
are all but eliminated.
The active shunt regulator and other regulators and ref-
erences on the chip are up to the design criteria. Specifi-
cally, the shunt regulator consumes negligible supply cur-
rent within the normal voltage range, with a very sharp
rise in current as the trigger point is exceeded. Previ-
ous systems use passive zener-based regulators, whose
gradual
I-V
characteristic provides insufficient overvolt-
age protection for low-voltage
CMOS
technologies.
The data link
to
the transceiver chip was tested by using
a Class
D
transmitter driving the inductive link, with a data
rate of 50kbitfs and
20%
ASK
modulation and Manch-
ester encoding. The data transfer functions according to
the specifications, and higher data rates can easily be sup-
ported with minor modifications.
4.
Conclusion
We have demonstrated a partitioning scheme for im-
planted sensor and actuator devices which places the sig-
nal processing functions where they are needed. By us-
ing this scheme, it is possible to provide better isolation
of weak biological signals from strong external distur-
bances, while simultaneously reducing the overall power
Figure
6.
Chip die microphotograph.
consumption by placing the transceiver closer to the exter-
nal interface. The communication between separate parts
of the system has been adapted to existing biocompatible
interconnect methods.
We have implemented a simple system consisting of a
transceiver and a single sensor chip, but the concept can
easily be extended to more general combinations of sen-
sors and actuators, in order to created a complete neural
stimulation system with a closed feedback loop.
[I]
G.
E.
Loeb and
R.
A. Peck, “Cuff electrodes for chronic
stimulation and recording
of
peripheral nerve activity”.
J.
Neuroscience Methods, 64
(1996),
95-103.
[2]
J.
H.
Nielsen and
T.
Lehmann,
“An
implantable CMOS
am-
plifier for nerve signals”,
Proc.
ZCECS
2001.
[3]
H.-T.
Ng and
D.
J.
Allstot, “CMOS current-steering logic
for
low-voltage mixed-signal integrated circuits”,
ZEEE
Truns.
VLSZ, 5
(1997),
pp.
301-308.
[4]
M.
Nardin and
K.
Najafi
“A
multichannel neuromuscular
microstimulator
with
bi-directional telemetry”,
1995
Con$
on Solid-state Sensors and Actuators,
pp. 59-62.
[5] D. Galbraith,
M.
Soma and
R.
White, “Wide-band efficient
inductjve transdermal power and data
link
with
coupling
insensitive gain”,
IEEE
Trans. BME, 34
(1
987), pp. 265-
215.
[6]
N.
de N. Donaldson and
T.
A.
Perkins. “Analysis
of
res-
onant coupled coils
in
the design of radio frequency tran-
scutaneous links”.
Med. and Bio. Eng. and Computing,
21
(1983),
pp. 612427.
[7] A. Djemouai,
M.
Sawan and M. Slamani,
“An
efficient
RF
power transfer and bidirectional data transmission to
im-
plantable electronic devices”,
Proc.
ISCAS
’99,
pp.
11-259.
[8]
P.
R.
Troyk
and
M.
A.
K. Schwan,
“Closed-loop class
E
transcutaneous power and data link
for
microimplants”
,
[9]
Z.
Tang
et
al. “Data transmission from an implantable
biotelemeter
by
load-shift
keying
using
a
circuit
configu-
ration modulator”.
IEEE
Trans. BME, 42
(1995), pp. 524.
Trans.
BME,
39
(1992), pp. 589-599.
400
Authorized licensed use limited to: Danmarks Tekniske Informationscenter. Downloaded on March 23,2010 at 09:13:27 EDT from IEEE Xplore. Restrictions apply.