180 IEEE TRANSACTIONS ON BIOMEDICAL CIRCUITS AND SYSTEMS, VOL. 13, NO.
1, FEBRUARY 2019
A 430-MHz Wirelessly Powered Implantable Pulse
Generator With Intensity/Rate Control and Sub-1 µA
Quiescent Current Consumption
Hongming Lyu , Member, IEEE, Parag Gad, Hui Zhong, V. Reggie Edgerton,
and Aydin Babakhani , Member, IEEE
Abstract—This work presents a miniaturized µW-level im-
plantable pulse generator (IPG) inductively powered at 430 MHz.
Notches are intentionally applied to the incident power, which are
replicated to precisely control the timing of the output pulses. Fab-
ricated in a 180-nm CMOS process, the concise circuitry occupies
a pad-included footprint of 850 µm × 450 µm and achieves a qui-
escent current consumption of 950 nA. To reduce the form factor,
401–457 MHz MedRadio-band is utilized to realize the induction
link. The finalized assembly achieves one of the smallest dimen-
sions (4.6 mm × 7.0 mm) for near-field IPGs with the Rx coil
size of 4.5 mm × 3.6 mm. Codesign of the rectifier and Rx coil
accommodates the possible resonant frequency drifts in biologi-
cal tissues. In the benchtop measurement, a 430-MHz Tx coil is
demonstrated to operate the IPG at 4.5 and 4 cm proximities in
the air and through water, respectively. An in vivo experiment has
been performed, in which the IPG was implanted on the hindlimb
muscle belly of an anesthetized rat with the connective tissue and
skin sutured. The electrical stimuli induced the isolated ankle flex-
ion at specific strengths and rates, and the experiment complies
with the specific absorption rate regulations. This work shows the
potential for applications requiring stringent form factors and high
sensitivities.
Index Terms—Electrical stimulation, energy-harvesting, im-
plantable medical device (IMD), implantable pulse generator
(IPG), inductive coupling, neuromuscular stimulation, voltage-
controlled stimulation, wireless power transfer.
I. INTRODUCTION
MPLANTABLE pulse generators (IPGs) have solved vari-
I ous critical clinical problems and improved the quality of
human life. Their applications include chronic pain relief [1],
motor function recovery for spinal cord injuries [2], the treat-
Manuscript received August 12, 2018; revised September 19, 2018 and Oc-
tober 21, 2018; accepted October 26, 2018. Date of publication November 9,
2018; date of current version January 25, 2019. This work was supported in part
by NSF and in part by Broccolli Foundation. This paper was recommended by
Associate Editor A. Demosthenous. (Corresponding author: Hongming Lyu.)
H. Lyu and A. Babakhani are with the Department of Electrical and Computer
Engineering, University of California at Los Angeles, Los Angeles, CA 90095
USA (e-mail:,hongminglyu@ucla.edu; aydinbabakhani@ucla.edu).
P. Gad, H. Zhong, and V. R. Edgerton are with the Departments of In-
tegrative Biology and Physiology, University of California at Los Angeles,
Los Angeles, CA 90095 USA (e-mail:,paraggad@ucla.edu; vzhong@ucla.edu;
vre@ucla.edu).
This paper has supplementary downloadable material available at
http://ieeexplore.ieee.org.
Color versions of one or more of the figures in this paper are available online
at http://ieeexplore.ieee.org.
Digital Object Identifier 10.1109/TBCAS.2018.2879357
1932-4545 © 2018 IEEE. Personal use is permitted, but republication/redistribution requires IEEE permission.
See http://www.ieee.org/publications standards/publications/rights/index.html for more information.
ment of gastroesophageal reflux disease [3], cardiac pacemaking
[4], and curing stress urinary incontinence [5], etc. Conventional
IPGs are bulky with the battery taking up most of the unit, and
the necessary leads are prone to cause various complications [1],
[4]. With this regard, recent research is interested in achieving
battery-less and leadless IPGs that can be directly implanted in
the specific anatomical region [5]–[8].
Most stimulation devices function in either current or voltage
modes. The current-controlled stimulation (CCS) provides pre-
cise current control irrelevant of the load impedance. However,
because the stimulator needs to comply with the worst-case elec-
trode/tissue impedance condition, the CCS renders the worse en-
ergy efficiency in most clinical settings. The voltage-controlled
stimulation (VCS) regulates the stimulus in the voltage domain
and renders an excellent energy efficiency. Due to this reason,
most existing commercially available IPGs are based on VCS
[9]. A physician identifying the appropriate range of stimulus
strength in advance and over time can eliminate the chance of
overstimulation [5].
Wireless power transfer is a substitute for the battery that pow-
ers implantable medical devices (IMDs). Aside from far/mid-
field coupling and ultrasonic transmission [10], [11], the near-
field inductive coupling is an attractive developing technology
[12], [13]. While prior induction-based IMDs were mostly pow-
ered at lower than 10s-MHz, the recent trend indicates that using
hundreds-MHz can substantially reduce the form factor of the
implants [14]–[17]. The medical device radiocommunications
(MedRadio) service, i.e., 401–406, 413–419, 426–432, 438–
444, and 451– 457 MHz, assigned by the federal communica-
tions commission has been used for the telemetry of IMDs [18],
[19]. However, its use in the power link has not been reported
to the authors’ knowledge. Unlike hundreds-MHz prior art that
adopts on-chip coils [14]–[16], this work seeks to implement
a miniaturized Rx coil on the PCB to minimize the cost. Also,
a discrete energy storage capacitor is regardless needed to be
assembled with the integrated circuitry.
In this paper, we propose a concise circuitry to realize an
energy-efficient voltage-controlled IPG with a quiescent (while
not stimulating) current consumption of 950 nA. Inductive cou-
pling at a MedRadio band achieves the wireless power link,
where notches are intentionally applied to precisely control
the width and rate of the output pulses in an analog manner.
The energy-harvesting frontend circuitry takes account of the
LYU et al.: 430-MHZ WIRELESSLY POWERED IPG WITH INTENSITY/RATE CONTROL AND SUB-1 μA QUIESCENT CURRENT CONSUMPTION
Fig. 1. Conceptual illustration of the in vivo experiment in which the IPG is
fully implanted and used to stimulate the animal’s hindlimb muscle.
Fig. 2. Circuitry overview. (a) Circuit architecture of the proposed IPG.
(b) Schematic of the Tx coil.
potential impacts of biological tissues. The finalized assembly
features an overall dimension of 4.6 mm × 7 mm with the Rx
coil size of 4.5 mm × 3.6 mm. The potential use of the IPG in
correcting the foot drop was verified in an in vivo study in which
the IPG was implanted at the hindlimb muscle (Tibialis Ante-
rior) belly of an anesthetized rat under the skin (Fig. 1). Isolated
contractions of the ankle joint were induced with controllable
rates and forces.
The remainder of the paper is organized as follows.
Section II elaborates on the circuit implementation with a fo-
cus on the design tradeoffs in the energy-harvesting frontend
circuitry. Section III presents the benchtop measurement and in
vivo experiment results. Section IV compares this work with
prior art and discusses future improvements. Section V con-
cludes the paper.
II. CIRCUIT IMPLEMENTATION
A. Circuit Overview
The systematic architecture of the IPG is shown in Fig. 2(a).
The magnetic field coupled to the Rx coil is rectified to gen-
erate VDD and charges an energy storage capacitor, CSTOR .
Notches are intentionally applied in the Tx signal which pre-
181
Fig. 3. The circuit schematic of the demodulator.
cisely controls the timing of the output stimulations as their
repetitions. The notch-based modulation scheme eliminates any
complex telemetry and minimizes the power consumption. As
the notches only constitute a negligible portion of the Tx power,
they do not degrade the efficiency of the power transfer link.
The VCS scheme is adopted in this work for better energy-
efficiency, in which VDD node is directly applied to the elec-
trode/tissue with a controllable pulse width. In replacement of
a low-dropout (LDO), a simplified output voltage regulator is
used to limit the amplitude of the output stimulations within a
specific range, which further reduces the static power consump-
tion. This regulator enables the notch-demodulation block only
when the supply voltage exceeds the lower tier. On the contrary,
when the supply voltage exceeds the higher tier, a discharge path
is enabled to rapidly discharge the excess incident charge. The
stimulations are delivered through a DC-block capacitor, CBCK ,
for charge-neutralization. A discharge resistor, RDIS , nulls the
accumulated charge on CBCK . A light-emitting diode (LED)
can be optionally included at the output.
The IPG is wirelessly powered and controlled by a custom
Tx coil with the diameter of approximately 3 cm (Fig. 2(b)). A
matching network ensures the impedance matching at approx-
imately 430 MHz, the resonant frequency of the Rx energy-
harvesting frontend.
B. Demodulator
The demodulator block is responsible for replicating the tim-
ing of the notch, as shown in Fig. 3. The conceptual waveforms
of the incident signal and the voltage of the critical nodes in
the demodulator are illustrated on the upper left. The circuit
consists of three source follower replicas. The high end, low
end, and transient envelope of the signal are denoted as VH ,
VL, and VENV , respectively. The VENV detection branch uses a
relatively small capacitor, CSM , while VH and VL are extracted
on larger capacitors with and without the AC input, respectively.
Because of the nonlinearity of the transistors’ transfer charac-
teristics, an AC swing applied on a constant gate bias generates
a larger source voltage. The average of VH and VL , VM , is ob-
tained through a resistive divider, which is thereafter compared
with VENV to reconstruct the timing of the notch. CSM and CLG
are selected to be 100 fF and 36 pF, respectively. As CSM <<
182 IEEE TRANSACTIONS ON BIOMEDICAL CIRCUITS AND SYSTEMS, VOL. 13, NO. 1, FEBRUARY 2019
Fig. 4. The circuit schematic of the output voltage regulator. (a) and (b) set
the high and low bars of the output amplitude, respectively. (c) generates the
voltage reference.
CLG , VM can be considered as constant so that the discharging
and charging of CSM determines the delays from the starting and
ending points, respectively. A smaller CSM renders a faster tran-
sient response yet suffers from a larger noise. The discharging
rate of CSM is independent of the amplitude of the Tx signal as
it is determined by the current source generated from a bandgap
reference block. The recovered timing signal is then sharpened
by a following buffer, as shown in the bottom figure. The buffer
only causes a sub-ns delay.
C. Output Voltage Regulator
Fractions of VDD are compared with a constant voltage ref-
erence, VREF , so that the amplitude can be regulated within a
specific range. Circuits in Fig. 4(a) and (b) determine the high
and low bars, respectively. When the supply voltage exceeds
19/12 of VREF , a discharge current path is enabled through
a 65 kΩ resistor, RD , which rapidly discharges the incident
power. On the contrary, when the amplitude is lower than 19/16
Fig. 5. The overall current consumption of the IC and that of the individual
blocks.
of VREF , OUT∗ node turns high, which disables the demodula-
tor in Fig. 3. The bandgap voltage reference circuit is shown in
Fig. 4(c). By tuning R1 and R2, VREF is designed to be 2.3 V,
which regulates the stimulation amplitude between 2.7 V and
3.6 V. This regulation scheme eliminates the widely used LDOs
which may turn to be the most static power-consuming block
in IMDs [14]. The voltage ladder can be further customized to
render a narrower window. It is also worth mentioning that in
the actual operation, an excessive Tx power tends to generate
pulses with the maximum amplitude.
The current consumption of the individual blocks is simu-
lated as shown in Fig. 5. With the onset of the demodulator at
around 2.7 V, the total current consumption of the IC, ITOT ,
features a rapid rise (due to the increase of IDEM ). When the
supply voltage reaches 3.6 V, the leakage path rapidly discharges
the incident power. Below that, the maximum ITOT is around
950 nA.
D. Energy-Harvesting Frontend
Modeling the input impedance of the rectifier as paralleled
R and C provides an intuitive insight into the rectifier design
for a resonant coupling system [20]. In the subthreshold region,
the input impedance of the rectifier is dominated by the gate
capacitances of the MOS transistors. On the contrary, as the
input voltage swing increases, transistors conduct more current
so that the input of the rectifier becomes more resistive.
The frontend resonator comprising the Rx coil, the rectifier,
and the demodulator can be modeled as the schematic shown
in Fig. 6. The Rx coil is modeled as the parallel configuration
of the inductance, LCOIL , the loss resistance, RCOIL , and the
parasitic capacitance, CCOIL . RREC and CREC represent the
input resistance and capacitance of the rectifier, respectively.
Similarly, RDEM and CDEM model the input characteristics of
the demodulator. However, as RDEM and CDEM are simulated
to be 1.2 MΩ and 4.7 fF, respectively, they can be omitted in
this work.
LYU et al.: 430-MHZ WIRELESSLY POWERED IPG WITH INTENSITY/RATE CONTROL AND SUB-1 μA QUIESCENT CURRENT CONSUMPTION 183

Fig. 6. The circuit model of the energy-harvesting frontend resonator.

Fig. 8. (a) Simplified model of the energy-harvesting frontend resonator.

(b) The circuit schematic of the Dickson rectifier.

Fig. 7. (a) The 3D model of the implemented Rx coil. (b) The picture of the
as-fabricated PCB incorporating the Rx coil.

The Rx coil dominantly determines the resonant frequency of

this resonator. Fig. 7 shows the 3D model and the as-fabricated
picture of the Rx coil. It resides on 0.5 mm thick Rogers 4350 B
substrate and features a five-turn design with two and three turns
on the top and bottom layers, respectively. The size of the Rx
coil is 4.5 mm × 3.6 mm. LCOIL is simulated to be 94.9 nH
taking account of all connected traces. As simulations indicate
CCOIL and RCOIL to be an order of magnitude larger than CREC
and RREC , respectively, the frontend resonator can be further
Fig. 9. (a) 3-dB bandwidth and (b) normalized Qη for different rectifier de-
simplified as in Fig. 8(a). The circuit schematic of the Dickson signs.
rectifier in this work is demonstrated in Fig. 8(b) [21]–[24].
Zero-threshold transistors are used to improve the conversion
efficiency [25]. with more stages and larger WG /LG render a larger 3 dB-
The design of the rectifier focuses on the tradeoff between bandwidth of the frontend resonator that can accommodate
the reception sensitivity and bandwidth. Assuming an ILOAD of larger dielectric medium variations (Fig. 9(a)). On the contrary,
5 μA, WG /LG ranging from 2.5 μm /0.5 μm to 20 μm /0.5 μm the fewer stages and the smaller WG /LG lead to a higher recep-
and the number of stages from 4 to 6 generate different reception tion sensitivity primarily owing to the increased quality factor,
bandwidths and sensitivities as shown in Fig. 9. Configurations Q (Fig. 9(b)). The reception sensitivity is compared as the
184 IEEE TRANSACTIONS ON BIOMEDICAL CIRCUITS AND SYSTEMS, VOL. 13, NO. 1, FEBRUARY 2019

Fig. 10. The simulated dependence of RR E C and CR E C on IL O A D . Fig. 11. The resonant frequency drift in muscle medium.

multiplication of Q and the intrinsic conversion efficiency, η, of

the rectifier [20]. The selected design in this work (WG /LG =
5 μm /0.5 μm, N = 5) renders a 24 MHz 3 dB-bandwidth and
an inherent conversion efficiency of 53% for the rectifier.
The selected rectifier design is further simulated to investigate
the impacts of ILOAD variations. With ILOAD varying from 1 μA
to 10 μA, CREC is remarkably stable at around 50 fF, which
verifies the stability of the resonant frequency of the energy-
harvesting frontend across a wide range of stimulation loads. On
the other hand, RREC decreases with ILOAD , which indicates an
increased reception sensitivity for a lighter load. The simulated
dependence of RREC and CREC on ILOAD is demonstrated in
Fig. 12. The co-design procedure for the Rx coil and the rectifier, which
Fig. 10. ensures the optimal performance at a specific MedRadio band.
The IPG assembly is encapsulated with epoxy. Therefore,
the frontend resonator is simulated within a 3 mm thick epoxy
and inside a 1.5 cm muscle cubic to provide an insight into
the potential impacts of the dielectric medium variations. The
simulation is performed with ANSYS and the result shows that
the muscle tissue causes a 9 MHz downward drift of the resonant
frequency as shown in Fig. 11. The selected rectifier design
succeeds in covering this drift within the 3-dB bandwidth.
Fig. 12 summarizes the procedure for the co-design of the
Rx coil and the rectifier targeting a specific MedRadio band.
The Rx coil plays the most dominant role in determining the
resonant frequency. The rectifier reaches the compromise be-
Fig. 13. Microscopic image of the fabricated IC.
tween the reception sensitivity and bandwidth according to the
specific load requirement. This process needs several iterations
of optimization to ensure a certain loaded resonant frequency. B. Tx Coil
The Tx coil features a single-turn design and is implemented
III. MEASUREMENT RESULTS
on an FR4 substrate, as shown in Fig. 15(a). The diameter and
A. Fabrication trace width are 29.7 mm and 1.52 mm, respectively. An L-
matching section ensures the impedance matching at 431 MHz
The IC was fabricated in TSMC 180 nm CMOS process with
as shown in the S11 measurement (Fig. 15(b)).
a pad-included area of 850 μm × 450 μm, as shown in Fig. 13. A
picture of the IPG assembly is shown in Fig. 14. Epoxy (Gorilla
C. IPG Output
4200101) was used to encapsulate the assembly and AWG 22
aluminum plated copper wire of about 5 mm was utilized as the The electrode impedance is modeled as a series combina-
electrodes for simplicity. tion of the tissue/solution resistance, RS , and the double-layer
LYU et al.: 430-MHZ WIRELESSLY POWERED IPG WITH INTENSITY/RATE CONTROL AND SUB-1 μA QUIESCENT CURRENT CONSUMPTION
Fig. 14. Picture of the as-fabricated IPG assembly in comparison with a US
dime.
Fig. 15. (a) Picture of the Tx coil and (b) its S11 according to measurement.
capacitance, CDL , according to previous works as shown in the
inset of Fig. 16 [26]–[28]. Two electrodes were immersed in the
phosphate buffered solution by approximately 5 mm. RS and
CDL were then characterized to be 1.2 kΩ and 0.6 μF, respec-
tively, with the Stanford Research System SR720 LCR Meter.
Due to the availability of the discrete components, RS of 1.15
kΩ and CDL of 0.6 μF in series were used as the load of the IPG.
A 6 μs notch was first applied to the Tx signal, which triggered
the output pulse as shown in Fig. 16. The monophasic waveform
185
Fig. 16. The output voltage waveform of the IPG in response to a 6 μs notch.
The inset shows the equivalent circuit model for the electrode.
has 4.7 μs and 1.4 μs delays compared to the starting and ending
points of the notch, respectively. Therefore, the duration of the
triggered stimulation is 3.3 μs shorter than that of the notch.
The spike at the onset of the pulse is an artifact due to parasitic
effects of the connection wire.
The voltage and corresponding current waveforms for the
96.7 μs and the 196.7 μs pulses are shown in Fig. 17. The
injected charge is temporarily accumulated on CDL so that
there appears a post-pulse voltage buildup. The voltage buildup
should not exceed the water delamination window, typically
about 1.4 V [29]. According to this constraint, the pulse width
should be kept below 300 μs. The current is obtained by record-
ing the voltage over the RS , which features an exponentially de-
caying waveform with the peak of approximately 3.2 mA (also
see Section III-E). A more comprehensive electrode model in-
cludes a charge transfer resistance, RCT , in parallel with CDL ,
which rapidly discharges the post-pulse potential in saline/tissue
[30]. In a typical case, RCT is around ten times as large as RS
[30]. With such RCT of 11 kΩ, the output voltage waveform
over multiple cycles is demonstrated in Fig. 17(e).
An LED can be optionally included at the output of the IPG to
indicate the occurrence of the output stimulation. In this work,
a green LED (APT1608LZGCK, Kingbright) is used. The IPG
is first tested in the air with the Tx power of 1 W. It shows the
maximum operating distance of 4.5 cm (Fig. 18(a)). The device
is then immersed in fresh water (εr = 80) at a 1.5 cm depth,
as the dielectric constant mimics that of the body. The Tx coil
operates the IPG at 2.5 cm above the water surface with a total
distance of 4 cm (Fig. 18(b)). The LED regulates the amplitude
of the output pulse at 3.1 V. 6.7 μs, Waveforms of 6.7 μs,
16.7 μs, and 26.7 μs pulses respectively triggered by 10 μs,
20 μs, and 30 μs notches are demonstrated in Fig. 19.
D. Animal Experiment
Selective activation of specific muscles with a miniaturized
implantable stimulator has been shown to correct foot drops
[31]. An in vivo experiment has been performed to test the use
186 IEEE TRANSACTIONS ON BIOMEDICAL CIRCUITS AND SYSTEMS, VOL. 13, NO. 1, FEBRUARY 2019

Fig. 17. Voltage (a, c) and the resulting current (b, d) waveforms for a 96.7 μs
pulse and a 196.7 μs pulse, respectively. (e) Three cycles of 96.7 μs pulses at
10 Hz rate.

of the IPG in neuromuscular stimulations. In the experiment, a

rat was initially anesthetized with urethane anesthesia (1.2 g/kg)
administered subcutaneously. An IPG device (w LED) was in-
serted into the muscle (Tibialis Anterior) belly with the two Fig. 18. The maximum-distance operations (a) in the air and (b) through water
electrodes about 2 mm apart. The device was secured in place with Tx power of 1 W.
with 4-0 Ethilon suture, as shown in the inset of Fig. 20(a). The
Tx coil was placed 3 cm above the hind limb with the source
ing the response to a stimulus, the motor output demonstrates
power of 1 W at 430 MHz. The connective tissue and skin were
minor variations due to the inherent variability in the nervous
sewn covering the device. The rat was placed on the back with
system. In addition, with each isometric contraction, the foot of
the knee joint secured using metallic screws. The toes were
the animal may be deflected, thus affecting the baseline force.
directly connected to a force transducer to measure isometric
The dependence of the induced force on the pulse width is
contractions, as shown in Fig. 20(a). Fig. 20(b) displays a closer
shown in Fig. 21(b). Peak to baseline force was calculated and
view of the implantation site. The force transducer was then
averaged for 10 pulses at each pulse width. The force monoton-
connected to a DAQ that digitizes and records the data (sam-
ically increases until a plateau for pulse widths above 100 μs.
pling frequency = 10 kHz). All procedures were in accordance
This non-linear relationship observed as a recruitment curve is
with the National Institute of Health Guide for the Care and
consistent with that observed previously [32]. The recruitment
Use of Laboratory Animals and were approved by the Animal
curve is a common strategy used for identifying the appropriate
Research Committee at UCLA.
stimulation parameters. A video recording of the induced ankle
The stimulation intensity was varied with each pulse width
flexion is presented in the Media information.
repeated at least 10 times to ensure reproducibility. The pulse
rate was fixed at 1 Hz in this experiment. A minimum of 2 min
E. Calculation of Charge Delivering
break was given between two pulse width cycles to account for
muscle fatigue. Transient recordings of the induced force with Calculation of the injected amount of charge provides
16.7 μs and 96.7 μs pulses are demonstrated in Fig. 21(a). Dur- an insight into the proper design of the electrodes for
LYU et al.: 430-MHZ WIRELESSLY POWERED IPG WITH INTENSITY/RATE CONTROL AND SUB-1 μA QUIESCENT CURRENT CONSUMPTION 187

Fig. 19. Output waveforms of an IPG with the LED loading the output.

Fig. 21. (a) Transient recordings of the induced force in response to 16.7 and
96.7 μs pulses. (b) The dependence of the induced force on the pulse width.

Fig. 20. (a) Animal experiment setup. The inset shows the implantation of
the IPG. (b) A closer view of the implantation site where the skin is sutured
covering the device.

voltage-controlled IPGs. Assuming the voltage buildup on

CBCK to be VX (VX typically much smaller than VDD), the
delivered amount of charge with each stimulation equals
 
ΔQch = (V DD − VX ) 1 − e−T P u l s e /R S C D L CD L (1)

where TPulse presents the pulse width. The amplitude of the

injected current exponentially decays as determined by the time
constant according to the electrode model shown in Fig. 16. In
the animal experiment, since the pulse amplitude is regulated
by the LED at around 3 V, 16.7 μs and 96.7 μs pulses deliver
approximately 0.04 μC and 0.23 μC charge, respectively. Mul-
tiplying ΔQch by the pulse rate, FPulse , the delivered amount of
charge in each second equals Fig. 22. Simulated 10-g average SAR when the Tx coil is placed at a distance
of 3 cm from a male right leg model in ANSYS.
Qch = FP u lse ΔQch (2)
188 IEEE TRANSACTIONS ON BIOMEDICAL CIRCUITS AND SYSTEMS, VOL. 13, NO. 1, FEBRUARY 2019

TABLE I
COMPARISON OF RECENTLY PUBLISHED BATTERY-LESS IPGS

a
The maximum TP u l s e varies according to the charge injection capacity of different electrodes.
b
FP u l s e limit of 300 kHz is when the pulse width approaches zero.
c
TP u l s e and FP u l s e need to suffice equation (1)–(3). The combination of TP u l s e and FP u l s e affects the selection of RD I S .
d
According to the simulation setup.
e
The thickness of the epoxy encapsulation is not well controlled and is excluded from the calculation.

Note that Qch is accumulated on CBCK . Therefore, the passive
discharging path should suffice the following relationship,
VX /RD I S > Qch
A smaller RDIS in the assembly will ensure a smaller VX that
does not evidently hamper the intensity of each stimulation. As
this work aims for a μW-level simulation load, RDIS is selected
to be 200 kΩ to ensure a minimum VX . CBCK is 47 μF in this
work. A relatively large CBCK helps to stabilize VX .
F. SAR Evaluation
The SAR evaluation is performed in ANSYS. Placing the Tx
coil at a 3 cm distance from the human leg model, the simulated
10-g averaged SAR features the maximum value of 1.645 W/kg
with the Tx power of 1 W, as shown in Fig. 22. The SAR
is well below the restrictions for localized exposure according
to IEEE Std C95.1-2005, i.e., the lower tier of 2 W/kg used for
general public and the higher tier of 10 W/kg used for controlled
environments, e.g., medical implant use.
IV. COMPARISONS AND FUTURE WORKS
A comparison with recently published miniaturized IPGs
is presented in Table I. Due to the elimination of the coil,
(3) ultrasound-based IPGs tend to have smaller form factors [7],
[8]. However, their operation typically requires the use of the
ultrasound gel. In addition, concerns were with its propaga-
tion through air-filled viscera such as the lung and bowel, and
obstructions such as bones. Passive circuits have also been in-
vestigated to realize energy-efficient IPGs [5], [33]. However,
they require sudden bursts of the Tx power, which are more
prone to violate the SAR regulations. To achieve a high recep-
tion sensitivity, the proposed IPG consumes one of the lowest
static powers among active circuitry-based works. The use of
MedRadio-band contributes to the miniaturized form factor of
the implant. Replacing the discrete components currently in
0603 SMD packages to 0201 ones can further reduce the overall
size by a large portion.
The proposed work can be improved with the following re-
gards. First, since this work uses the passive charge-balancing
DC-block, an offset voltage would build up over the electrode-
tissue interface in an intensive stimulation setting, which needs
to be analyzed and monitored [34]. Second, while this work aims
LYU et al.: 430-MHZ WIRELESSLY POWERED IPG WITH INTENSITY/RATE CONTROL AND SUB-1 μA QUIESCENT CURRENT CONSUMPTION
to demonstrate the operation with a relatively long Tx-Rx sepa-
ration, future works can investigate a wearable Tx antenna that
is placed at a close proximity and with much-reduced power for
subcutaneous applications. Third, long-term applications should
consider the use of biocompatible electrode materials such as
iridium oxide, carbon, and noble metals [29], [35].
V. CONCLUSION
In this work, we propose a battery-less IPG based upon a
concise circuitry and achieving an mm-scale form factor. A
novel notch-modulation scheme precisely controls the rate and
intensity of the output pulses. 401–457 MHz MedRadio-band
is used to reduce the size of the inductive Rx coil. The finalized
assembly possesses a form factor of 4.6 mm × 7 mm.
As the circuitry only consumes about 3 μW static power, this
work easily complies with the SAR regulations. With the Tx
power of 1 W, benchtop measurements demonstrate the max-
imum operating distances of 4.5 cm and 4 cm in the air and
through fresh water, respectively. To test its potential in correct-
ing foot drops, the IPG was implanted at the hind limb muscle
belly of a rat with the connective tissue and skin sutured cover-
ing device. Ankle flexion forces with controllable strengths and
rates were successfully induced.
The quiescent power consumption of the proposed work could
be one of the lowest among active circuitry-based IPGs. After
further reducing the form factor, this work shows the potential
for applications requiring extremely small form factors and high
sensitivities.
REFERENCES
[1] T. R. Deer et al., “The appropriate use of neurostimulation: Avoidance and
treatment of complications of neurostimulation therapies for the treatment
of chronic pain,” Neuromodulation, Technol. Neural Interface, vol. 17,
no. 6, pp. 571–598, 2014.
[2] Y.-K. Lo et al., “A fully integrated wireless SoC for motor function recov-
ery after spinal cord injury,” IEEE Trans. Biomed. Circuits Syst., vol. 11,
no. 3, pp. 497–509, Jun. 2017.
[3] L. Rodriguez et al., “Long-term results of electrical stimulation of the
lower esophageal sphincter for the treatment of gastroesophageal reflux
disease,” Endoscopy, vol. 45, no. 8, pp. 595–604, 2013.
[4] J. B. van Rees, M. K. de Bie, J. Thijssen, C. J. W. Borleffs, M. J. Schalij,
and L. van Erven, “Implantation-related complications of implantable
cardioverter-defibrillators and cardiac resynchronization therapy devices:
A systematic review of randomized clinical trials,” J. Amer. College Car-
diol., vol. 58, no. 10, pp. 995–1000, 2011.
[5] X. Huang, K. Zheng, S. Kohan, P. M. Denprasert, L. Liao, and G. E.
Loeb, “Neurostimulation strategy for stress urinary incontinence,” IEEE
Trans. Neural Syst. Rehabil. Eng., vol. 25, no. 7, pp. 1068–1078, Jul.
2017.
[6] Q. Xu, D. Hu, B. Duan, and J. He, “A fully implantable stimulator with
wireless power and data transmission for experimental investigation of
epidural spinal cord stimulation,” IEEE Trans. Neural Syst. Rehabil. Eng.,
vol. 23, no. 4, pp. 683–692, Jul. 2015.
[7] J. Charthad et al., “A mm-sized wireless implantable device for electrical
stimulation of peripheral nerves,” IEEE Trans. Biomed. Circuits Syst.,
vol. 12, no. 2, pp. 257–270, Apr. 2018.
[8] B. C. Johnson et al., “StimDust: A 6.5 mm3 , wireless ultrasonic peripheral
nerve stimulator with 82% peak chip efficiency,” in Proc. 2018 IEEE
Custom Integr. Circuits Conf., 2018, pp. 1–4.
[9] M. Niemann, G. H. Schneider, A. Kühn, P. Vajkoczy, and K. Faust,
“Longevity of implantable pulse generators in bilateral deep brain stimu-
lation for movement disorders,” Neuromodulation, Technol. Neural Inter-
face, vol. 21, no. 6, pp. 597–603, 2018.
189
[10] J. Charthad, M. J. Weber, T. C. Chang, and A. Arbabian, “A mm-sized
implantable medical device (IMD) with ultrasonic power transfer and a
hybrid bi-directional data link,” IEEE J. Solid-State Circuits, vol. 50, no. 8,
pp. 1741–1753, Aug. 2015.
[11] J. S. Ho et al., “Wireless power transfer to deep-tissue microim-
plants,” Proc. Nat. Acad. Sci. USA, vol. 111, no. 22, pp. 7974–7979,
2014.
[12] M. A. Hannan, S. Mutashar, S. A. Samad, and A. Hussain, “Energy
harvesting for the implantable biomedical devices: Issues and challenges,”
Biomed. Eng. Online, vol. 13, no. 1, p. 79, 2014. [Online]. Available:
https://doi.org/10.1186/1475-925X-13-79.
[13] A. M. Jawad, R. Nordin, S. K. Gharghan, H. M. Jawad, and M. Ismail,
“Opportunities and challenges for near-field wireless power transfer: A
review,” Energies, vol. 10, no. 7, p. 1022, 2017. [Online]. Available:
https://doi.org/10.3390/en10071022.
[14] C. Kim, S. Ha, J. Park, A. Akinin, P. P. Mercier, and G. Cauwenberghs, “A
144-MHz fully integrated resonant regulating rectifier with hybrid pulse
modulation for mm-sized implants,” IEEE J. Solid-State Circuits, vol. 52,
no. 11, pp. 3043–3055, Nov. 2017.
[15] A. Agarwal et al., “A 4 μW, ADPLL-based implantable amperometric
biosensor in 65 nm CMOS,” in Proc. 2017 Symp. VLSI Circuits, 2017,
pp. C108–C109.
[16] H. Jiang, X. Zhou, S. Kulkarni, M. Uranian, R. Seenivasan, and D. A.
Hall, “A sub-1 μW multiparameter injectable BioMote for continuous
alcohol monitoring,” in Proc. 2018 IEEE Custom Integr. Circuits Conf.,
2018, pp. 1–4.
[17] D. Ahn and M. Ghovanloo, “Optimal design of wireless power transmis-
sion links for millimeter-sized biomedical implants,” IEEE Trans. Biomed.
Circuits Syst., vol. 10, no. 1, pp. 125–137, Feb. 2016.
[18] J. Pandey and B. P. Otis, “A sub-100 μW MICS/ISM band transmitter
based on injection-locking and frequency multiplication,” IEEE J. Solid-
State Circuits, vol. 46, no. 5, pp. 1049–1058, May 2011.
[19] Y.-H. Liu, L.-G. Chen, C.-Y. Lin, and T.-H. Lin, “A 650-pJ/bit medradio
transmitter with an FIR-embedded phase modulator for medical micro-
power networks (MMNs),” IEEE Trans. Circuits Syst. I, Reg. Papers,
vol. 60, no. 12, pp. 3279–3288, Dec. 2013.
[20] H. Lyu, J. Wang, J.-H. La, J. M. Chung, and A. Babakhani, “An energy-
efficient wirelessly powered millimeter-scale neurostimulator implant
based on systematic codesign of an inductive loop antenna and a custom
rectifier,” IEEE Trans. Biomed. Circuits Syst., vol. 12, no. 5, pp. 1131–
1143, Oct. 2018.
[21] J. F. Dickson, “On-chip high-voltage generation in MNOS integrated cir-
cuits using an improved voltage multiplier technique,” IEEE J. Solid-State
Circuits, vol. SSC-11, no. 3, pp. 374–378, Jun. 1976.
[22] Y.-K. Teh, F. Mohd-Yasin, F. Choong, M. I. Reaz, and A. V. Kordesch,
“Design and analysis of UHF micropower CMOS DTMOST rectifiers,”
IEEE Trans. Circuits Syst. II, Express Briefs, vol. 56, no. 2, pp. 122–126,
Feb. 2009.
[23] J. Yi, W.-H. Ki, and C.-Y. Tsui, “Analysis and design strategy of UHF
micro-power CMOS rectifiers for micro-sensor and RFID applications,”
IEEE Trans. Circuits Syst. I, Reg. Papers, vol. 54, no. 1, pp. 153–166,
Jan. 2007.
[24] J. Charthad, N. Dolatsha, A. Rekhi, and A. Arbabian, “System-level analy-
sis of far-field radio frequency power delivery for mm-sized sensor nodes,”
IEEE Trans. Circuits Syst. I, Reg. Papers, vol. 63, no. 2, pp. 300–311,
Feb. 2016.
[25] P. T. Theilmann, C. D. Presti, D. J. Kelly, and P. M. Asbeck, “A μW
complementary bridge rectifier with near zero turn-on voltage in SOS
CMOS for wireless power supplies,” IEEE Trans. Circuits Syst. I, Reg.
Papers, vol. 59, no. 9, pp. 2111–2124, Sep. 2012.
[26] S. K. Kelly and J. L. Wyatt, “A power-efficient neural tissue stimulator
with energy recovery,” IEEE Trans. Biomed. Circuits Syst., vol. 5, no. 1,
pp. 20–29, Feb. 2011.
[27] S. K. Arfin and R. Sarpeshkar, “An energy-efficient, adiabatic electrode
stimulator with inductive energy recycling and feedback current regula-
tion,” IEEE Trans. Biomed. Circuits Syst., vol. 6, no. 1, pp. 1–14, Feb.
2012.
[28] J. E. B. Randles, “Kinetics of rapid electrode reactions,” Discuss. Faraday
Soc., vol. 1, pp. 11–19, 1947.
[29] S. F. Cogan, “Neural stimulation and recording electrodes,” Annu. Rev.
Biomed. Eng., vol. 10, pp. 275–309, 2008.
[30] Y.-K. Lo, C.-W. Chang, and W. Liu, “Bio-impedance characterization
technique with implantable neural stimulator using biphasic current stim-
ulus,” in Proc. 2014 36th Annu. Int. Conf. IEEE Eng. Med. Biol. Soc.,
2014, pp. 474–477.
190 IEEE TRANSACTIONS ON BIOMEDICAL CIRCUITS AND SYSTEMS, VOL. 13, NO. 1, FEBRUARY 2019
[31] D. Weber et al., “Functional electrical stimulation using microstimulators
to correct foot drop: A case study,” Can. J. Physiol. Pharmacol., vol. 82,
nos. 8/9, pp. 784–792, 2004.
[32] D. G. Sayenko et al., “Spinal segment-specific transcutaneous stimulation
differentially shapes activation pattern among motor pools in humans,” J.
Appl. Physiol., vol. 118, no. 11, pp. 1364–1374, 2015.
[33] K. L. Montgomery et al., “Wirelessly powered, fully internal optogenetics
for brain, spinal and peripheral circuits in mice,” Nature Methods, vol. 12,
no. 10, pp. 969–974, 2015.
[34] M. N. van Dongen and W. A. Serdijn, “Does a coupling capacitor enhance
the charge balance during neural stimulation? An empirical study,” Med.
Biol. Eng. Comput., vol. 54, no. 1, pp. 93–101, 2016.
[35] Y. Jia et al., “A mm-sized free-floating wirelessly powered implantable
optical stimulating system-on-a-chip,” in Proc. 2018 IEEE Int. Solid-State
Circuits Conf., 2018, pp. 468–470.
[36] Y. Lu, H. Lyu, A. G. Richardson, T. H. Lucas, and D. Kuzum, “Flexible
neural electrode array based-on porous graphene for cortical microstimu-
lation and sensing,” Sci. Rep., vol. 6, Sep. 19, 2016, Art. no. 33526.
Hongming Lyu (M’18) received the B.S. degree from
the University of Electronic Science and Technology
of China, Chengdu, China, in 2010, and the Ph.D.
degree from Tsinghua University, Beijing, China,
in 2015, both in electrical engineering. From 2015
to 2016, he was a Postdoctoral Research Fellow
with UCSD, La Jolla, CA, where he researched on
graphene-based neural interfaces. Since 2016, he has
been a Postdoctoral Research Fellow with Rice Uni-
versity, Houston, TX, and then with UCLA, Los An-
geles, CA. His major research interests include low-
power integrated circuits and wireless technologies for biomedical implants.
Parag Gad received the Bachelor’s in Engineering
degree in biomedical engineering from Mumbai Uni-
versity, Mumbai, India, in 2008. In the fall of 2008,
he started graduate school at UCLA working under
Dr. V Reggie Edgerton toward the Master’s and Ph.D.
degrees in biomedical engineering. He received the
Master’s and Ph.D. degrees in 2010 and 2013, respec-
tively, focusing on locomotion rehabilitation and im-
provement in bladder function using epidural spinal
cord stimulation after paralysis. He has been a Re-
searcher with the Department of Integrative Biology
and Physiology, UCLA, under Dr. V Reggie Edgerton’s guidance developing
tools and technologies to activate spinal neural networks to enable physiological
function after paralysis.
Hui Zhong received the Medical degree from Heibei
Medical University in 1983. From 1983 to 1991, she
was an Attending Physician and Researcher with the
Department of Emergency Medicine, Taian Central
Hospital, China. In 1993, she joined Dr. Edgerton’s
Neuromuscular Research Laboratory, University of
California at Los Angeles, and is currently a Project
Scientist.
V. Reggie Edgerton received the B.S. degree from
East Carolina University, the Master’s degree from
the University of Iowa, and the Ph.D. degree in exer-
cise physiology from Michigan State University. He
is currently the Director of the Neuromuscular Re-
search Laboratory and a Distinguished Professor of
the Departments of Integrative Biology and Physiol-
ogy, Neurobiology and Neurosurgery. He has been
teaching and conducting research at UCLA for more
than 50 years. His research is focused on how the neu-
ral networks in the lumbar spinal cord of mammals,
including humans, regain control of standing, stepping and voluntary control of
fine movements after paralysis, and how can these motor functions be modi-
fied by chronically imposing activity-dependent interventions after spinal cord
injury.
Aydin Babakhani (M’08) received the B.S. degree
from the Sharif University of Technology in 2003,
and the M.S. and Ph.D. degrees from Caltech in 2005
and 2008, respectively, both in electrical engineering.
He is an Associate Professor with the Department
of Electrical and Computer Engineering, UCLA, and
the Director of UCLA Integrated Sensors Laboratory.
Prior to joining UCLA, he was an Associate Professor
with the Department of Electrical and Computer En-
gineering, Rice University, and the Director of Rice
Integrated Systems and Circuits Laboratory. He was
a Louis Owen Junior Chair Assistant Professor (2016–2017) and an Assistant
Professor of ECE (2011–2016) with Rice University. He is a Member of DARPA
Microsystems Exploratory Council and a Co-Founder of MicroSilicon, Inc. He
was a Postdoctoral Scholar with Caltech in 2009 and a Research Scientist with
IBM Thomas J. Watson Research Center in 2010. He has authored or coauthored
more than 85 papers in peer-reviewed journals and conference proceedings as
well as 21 issued or pending patents.
Dr. Babakhani has been awarded multiple best paper awards, including the
Best Paper Award at the IEEE SiRF Conference in 2016, the Best Paper Award
at the IEEE RWS Symposium in 2015, the Best Paper Award at the IEEE IMS
Symposium in 2014, and the 2nd-place in the Best Paper Awards at the IEEE
APS Symposium 2016 and IEEE IMS Symposium 2016. His research is sup-
ported by NSF, DARPA, AFOSR, ONR, the W. M. Keck Foundation, SRC, and
more than 10 companies. He received a prestigious NSF CAREER award in
2015, an Innovation Award from Northrop Grumman in 2014, and a DARPA
Young Faculty Award in 2012. He also received the Caltech Electrical Engi-
neering Department’s Charles Wilts Best PhD Thesis Prize for his work titled
“Near-Field Direct Antenna Modulation.” He was the recipient of the Microwave
Graduate Fellowship in 2007, the Grand Prize in the Stanford-Berkeley-Caltech
Innovators Challenge in 2006, the Analog Devices, Inc., Outstanding Student
Designer Award in 2005, as well as a Caltech Special Institute Fellowship, and
an Atwood Fellowship in 2003. He was also the Gold Medal Winner at both
the National Physics Competition in 1998 and the 30th International Physics
Olympiad in Padova, Italy, in 1999.