7470 IEEE TRANSACTIONS ON POWER ELECTRONICS, VOL. 37, NO.

6, JUNE 2022

Analysis of a Dual-Transduction Receiver for

Electrodynamic Wireless Power Transfer
Miah A. Halim , Adrian A. Rendon-Hernandez , Spencer E. Smith ,
and David P. Arnold , Senior Member, IEEE

Abstract—In this article, we present the characterization and (EWPT) (a.k.a. “magneto-mechano-electric”) systems leverage
an experimentally validated electromechanical model of a dual- low-frequency magnetic fields (<1 kHz) to facilitate wireless
transduction receiver for electrodynamic wireless power transfer power delivery to compact electromechanical receivers, even in
(EWPT) system. This dual-transduction EWPT receiver contains the presence of electrically conductive objects [7]. Additionally,
two unimorph-type, series-connected piezoelectric transducers and human exposure to high-frequency magnetic fields is tightly
an electrodynamic transducer within a compact footprint and
low-profile design. The receiver makes simultaneous use of both
regulated [8]–[10], and therefore low-frequency EWPT systems
piezoelectric and electrodynamic transducers to generate electrical may be more amenable for wearable and implantable biomedical
power while operating at its torsion mode mechanical resonance devices.
at ∼744 Hz. The electromechanical system behavior and output In principle, the low-frequency magnetic field generated by a
performance under low-frequency, magnetic near-fields are an- transmitter excites a mechanical motion of a permanent magnet
alyzed by establishing an equivalent lumped-element equivalent in the EWPT receiver, which is then converted to electrical
electrical circuit model. A prototype device is fabricated, assem- energy via one or more electromechanical transduction schemes.
bled, characterized, and the experimental results are compared Commonly used transduction schemes include piezoelectric
with the model predictions under various excitation and loading (PE) and electrodynamic (ED) (interaction between permanent
conditions. A maximum of 49 µW average power is generated magnet and a coil) [11], [12]. In order to optimize the perfor-
by the prototype under 120 µTrms , which corresponds to 0.54
mW·cm–3 power density and 37 mW·cm–3 ·mT–2 normalized power
mance of an EWPT receiver under any given magnetic field
density. This chip-sized (0.09 cm3 ) design offers an innovative and and to maximize the power transfer efficiency of the EWPT
volume-efficient architecture for application in wirelessly charging system, multienergy-domain lumped-element modeling is a
wearable and implantable medical devices. widely used and effective system modeling approach [13], [14].
For instance, design, modeling, and experimental validation of
Index Terms—Dual-transduction, electrodynamic, electromech- EWPT systems using PE and ED receivers have been reported
anical coupling, piezoelectric, torsional resonance, wearables and to parameterize the corresponding system and to predict their
bio-implants, wireless power transfer (WPT).
output performances [15]–[17].
It is well known that PE transducers generally have high
I. INTRODUCTION electrical output impedance and generally produce higher
voltage but lower current. ED transducers, on the other hand,
LECTROMECHANICAL transducers (e.g., piezoelectric,
E electromagnetic/electrodynamic, electrostatic, etc.) have
long been studied for kinetic energy harvesting systems to power
have lower output impedance and generate lower voltage and
higher current. For example, in [18], for a magnetic field
strength of ∼0.4 mTrms , the induced voltage for the reported
or recharge batteries in modern electronic systems such as wear- PE EWPT receiver was higher (∼9.2 Vrms ) but produced a
able and implantable biomedical devices [1], [2]. However, due lower average power (∼0.13 mW). Under the same operating
to the intermittency of ambient energy and relatively low charg- condition, an ED receiver [12] generated relatively low voltage
ing rates (often below 1 μW), interest in wireless power transfer (∼2.1 Vrms ), but delivered a few milliwatts of average power
(WPT) methods has been growing in order to provide more (∼1.8 mW). Generally, higher voltages are desired to facilitate
deterministic control and to facilitate higher recharge rates [3], efficient downstream power management circuits, whereas
[4]. Compared with the conventional inductive and magnetic res- significant power (the product of voltage and current) is
onance near-field WPT schemes [5], [6], electrodynamic WPT required for any practical electronic load application such as
wearable and/or implantable medical devices. As one example,
Manuscript received April 27, 2021; revised July 7, 2021, September 10, 2021, the comparatively larger voltages produced by the PE transducer
and November 10, 2021; accepted December 31, 2021. Date of publication could be used to power active power electronics, whereas the
January 6, 2022; date of current version February 18, 2022. This work was larger current waveforms produced by the ED transducer could
supported by the IoT4Ag Engineering Research Center funded by the National
Science Foundation (NSF) under NSF Cooperative Agreement EEC-1941529.
provide the primary power path. The ED voltages are too small
Recommended for publication by Associate Editor O. C. Onar. (Corresponding (often << 1 V) for passive rectification as well as to energize any
author: Miah A. Halim.) active rectification circuitry. In order to achieve the “best of both
The authors are with Interdisciplinary Microsystems Group, University of worlds,” our group has recently reported the design, fabrication,
Florida, Gainesville, FL 32611 USA (e-mail: md.miah@ufl.edu; arendonher- and experimental characterization of a dual-transduction EWPT
nandez@ufl.edu; smithspencer12@ufl.edu; darnold@ufl.edu).
Color versions of one or more figures in this article are available at receiver with a volume-efficient and low-profile design that
https://doi.org/10.1109/TPEL.2022.3140777. makes use of both PE and ED transducers [19]. The experimental
Digital Object Identifier 10.1109/TPEL.2022.3140777 results indicated that the dual-transduction scheme increased

0885-8993 © 2022 IEEE. Personal use is permitted, but republication/redistribution requires IEEE permission.
See https://www.ieee.org/publications/rights/index.html for more information.

Authorized licensed use limited to: College of Engineering - THIRUVANANTHAPURAM. Downloaded on August 22,2023 at 10:33:22 UTC from IEEE Xplore. Restrictions apply.
 HALIM et al.: ANALYSIS OF A DUAL-TRANSDUCTION RECEIVER FOR ELECTRODYNAMIC WIRELESS POWER TRANSFER 7471

Fig. 1. Schematic drawings of the dual-transduction receiver. (a) Top view.

(b) Cross-sectional view with interacting magnetic field lines.

the total power (0.52 mW) and hence the power density
(5.8 mW·cm–1 ) over the individual PE and ED counterparts of
the receiver.
This article explores the electromechanical system behav- Fig. 2. FEA analysis. (a) First torsional mode resonance. (b) Frequency
ior of the previously reported dual-transduction EWPT re- response for torsional rotation angle under 50 μTrms field with ζ = 0.0056
(Q = 90) for various loading conditions.
ceiver, whereas both transducers operate simultaneously. Finite-
element analysis (FEA) is used to study its torsional mode
operation. Next, an equivalent lumped-element model (LEM)
of the coupled system is developed in order to analyze the The modal analysis of the receiver is studied by a three-
system performance under various loading and input conditions. dimensional FEA simulation using COMSOL Multiphysics, as
Finally, the model is validated by experimentally characterizing illustrated in Fig. 2. To simulate the bond layers (between the
a fabricated and assembled dual-transduction EWPT receiver magnet, spacer, suspension, and between piezo-element and
prototype. suspension arm), a 20 μm thick elastic layer with Young’s
modulus E = 2 GPa and Poisson’s ratio ν = 0.25 is used. As
shown in Fig. 2(a), the first mode of vibration occurs at 744.2 Hz,
II. DESIGN AND OPERATION
which is a torsional rotation about the diagonal axis a–a . Also,
Schematic drawings of the dual-transduction receiver are a frequency-domain FEA simulation, as shown in Fig. 2(b),
illustrated in Fig. 1. The receiver comprises an oscillating is carried out to investigate the frequency response (torsional
structure that combines both PE and ED transducers. The PE rotation) for various loading conditions of the transducers, e.g.,
transducer is formed by attaching two piezoelectric elements 1) both PE and ED open circuit, 2) PE open circuit with ED
to the clamped arms of a meandering suspension in a way so short, 3) PE short circuit with ED open, and 4) both PE and ED
that the two unimorphs are in series electrical connection. The loaded with resistive loads of 600 kΩ and 160 Ω, respectively
ED transducer is formed by attaching a laterally magnetized (experimentally obtained optimum load resistance values). The
permanent magnet to the center platform of the meandering simulation uses a 50 μTrms field and mechanical damping
suspension via a spacer and a rectangular coil attached to the ratio of ζ = 0.0056 (corresponds to measured quality factor
anchor that surrounds the magnet, both on the side opposite to (Q-factor) of 90). From the frequency response in Fig. 2(b),
the piezoelectric elements. The spacer provides clearance for it is observed that the PE transducer loading condition has an
the magnet while it oscillates. The overall dimensions of the influence on the resonant frequency of the system, whereas that
receiver are 7.6 × 7.6 × 1.65 mm3 , where the low-profile form of the ED transducer dominates the torsional rotation amplitude
factor facilitates future integration (e.g., via printed circuit board of the system. However, the maximum torsional rotation angle
(PCB) assembly) with other electronic components. of 0.27° is reached for PE short-circuited with ED open, and the
When the receiver is subjected to a time-varying magnetic corresponding (short-circuit) resonant frequency is 741.9 Hz.
near-field of desired frequency and amplitude, a torque is in- When both are open-circuited, the maximum torsional rotation
duced on the receiver magnet. This architecture allows the angle and the corresponding (open-circuit) resonant frequency
magnet to oscillate torsionally resulting in a dynamic stress are 0.27° and 744.2 Hz, respectively. As for the loaded condition
generated on the piezoelectric elements of the PE transducer, for both PE and ED, it is found that the maximum torsional
which, in turn, produces electricity via direct piezoelectric ef- rotation angle and the (loaded-circuit) resonant frequency are
fect. Simultaneously, the oscillation of the magnet induces an 0.22° and 743.4 Hz, respectively. For the ED short-circuited with
electromotive force (EMF) in the receiver coil by means of PE open, decrease in torsional rotation angle (0.23° at 744.2 Hz)
electrodynamic transduction. For both PE and ED transducers, is observed, which is attributed due to the Lorentz force in the
maximum voltage and power are generated at the torsional ED transducer. It is evident that the torsional rotation of the
resonance of the suspension structure with the receiver oriented magnet is very small, which limits the induced EMF voltage in
perpendicular to the interacting magnetic fields [as illustrated in the ED transducer. An optimized piezo thickness would increase
Fig. 1(b)]. the rotational angle (by achieving optimal equivalent suspension

Authorized licensed use limited to: College of Engineering - THIRUVANANTHAPURAM. Downloaded on August 22,2023 at 10:33:22 UTC from IEEE Xplore. Restrictions apply.
 7472 IEEE TRANSACTIONS ON POWER ELECTRONICS, VOL. 37, NO. 6, JUNE 2022

Fig. 3. Lumped element equivalent electrical circuit representation of the EWPT system with the dual-transduction receiver.

stiffness) and hence, the voltage/power generated by both PE and (tanδ) has negligible influence on the overall impedance and
ED transducers [15]. hence is neglected in the model [15].

III. SYSTEM MODELING B. System Parameters

In principle, each lumped-element parameter of the proposed
A. Lumped Element Equivalent Circuit system can be derived as a priori from relevant geometric
Fig. 3 shows the lumped-element equivalent electrical circuit dimensions and material properties. Most of the system pa-
of the proposed EWPT system indicating the electrical behavior rameters can also be determined via appropriate experimental
of a transmitter, the mechanical (rotational mechanics) and tests. Since some parameters may be easier to calculate, whereas
electrical behavior of the dual-transduction receiver, and the some parameters may be easier to measure/infer, a combination
electrical loads ZL−PE and ZL−ED connected, respectively, to of approaches such as analytical expressions, COMSOL FEA
the PE and ED transducers of the receiver. Phasor voltages, results, and experiments will be used.
currents, torques, and angular velocities are the primary sys- For analysis and experimental purposes, a Helmholtz coil pair
tem variables. Here, a power source with source resistance RS is considered as the source of the magnetic field. It produces a
supplies alternating voltage VS and current IS to the transmitter, uniform magnetic field at the center whose amplitude is con-
represented by an electrical resistance RT and inductance LT . trolled by the supply current IS . The amplitude of the on-axis
The mechanical oscillator is represented by torsional damping [refer to Fig. 1(b)] magnetic field is calculated as
coefficient b, magnet-mass moment of inertia J and short-circuit
torsional spring stiffness k. The coupling/linkage between the 8 μ0 N
Bz = √ IS (1)
transmitter and the receiver is achieved through electrodynamic 5 5 R
transduction represented by a gyrator with transduction co-
efficient KT . The clamped electrical capacitance of the PE where μ0 ( = 4π × 10–7 H·m–1 ) is the permeability of free space,
transducer is represented by a capacitor C0 and the receiver N is the number of turns in each coil, and R is the distance
coil of the ED transducer is represented by a resistance RR between the two coils. Once the receiver is subjected to the
and an inductance LR . Finally, these electrical elements of the magnetic field Bz , the torque induced on the receiver magnet
PE and ED transducers are coupled to the mechanical elements can be determined as
via a transformer of turn ratio ΓP and a gyrator having ED Br
transduction coefficient KR , respectively. τmag = vmag Bz (2)
While the system operates, the magnetic fields produced μ0
by the transmitter (not explicit in the equivalent circuit, but where Br is the remanence of the magnetic material and vmag
discussed later) induces torque τmag on the receiver magnet is the volume of the magnet. Note that the magnetization axis of
and allows the mechanical system to oscillate torsionally at an the receiver magnet is oriented perpendicular to the interacting
angular velocity θ̇. Therefore, an equivalent inertial force/ torque magnetic field (along z-axis) in order to maximize the torque on
τPE acts on the PE transducer elements that develops strain in the receiver magnet.
the piezo-materials, induces charges across the piezo-capacitor According to the fundamentals of electromechanical trans-
C0 and hence, a voltage VPE is generated. Simultaneously, duction principle and the electrical and mechanical equations of
the relative motion between the receiver magnet and the coil equilibrium [13], [14]
develops an equivalent electrodynamic torque τED in the ED
transducer, which in turn induces an EMF voltage VED . It is to be τmag
KT = (3)
noted that the dielectric loss tangent of the piezoelectric material IS

Authorized licensed use limited to: College of Engineering - THIRUVANANTHAPURAM. Downloaded on August 22,2023 at 10:33:22 UTC from IEEE Xplore. Restrictions apply.
 HALIM et al.: ANALYSIS OF A DUAL-TRANSDUCTION RECEIVER FOR ELECTRODYNAMIC WIRELESS POWER TRANSFER 7473

ZPE and ZED represent equivalent impedances of the PE and ED

transducers with corresponding load resistances, respectively.
Now, the voltage and power generated by each transducer of the
dual-transduction receiver can easily be determined by standard
ac circuit analysis.
The angular velocity of the receiver magnet due to the driving
torque τmag is given by
τmag τmag
θ̇ = =  (8)
Zm b + jωJ + k
+ Γ2P ZPE +
KR2

jω ZED

1 RL−PE
where ZPE = jωC 0
||RL−PE = 1+jωC 0 RL−PE
and ZED = RR +
RL−ED . According to lossless two-port transducer design prin-
ciple [13], [14], the voltages simultaneously generated across the
load resistances of the PE and ED transducers are, respectively
Fig. 4. Simplified equivalent electrical circuit of the dual-transduction receiver  
with resistive loads connected to each transducer. k
τmag − b + jωJ + jω θ̇ − τED
VL−PE = (9)
ΓP
VED RL−ED
KR = (4) VL−ED = VED (10)
θ̇ RR + RL−ED
 
k = 1 − κ 2 k0 (5) where τED = KR IL−ED = KR RR +R VED
and VED = KR θ̇.
  L−ED

C0 = 1 − κ2 C (6) Substituting and rearranging (8)–(10), the voltage equations are

√ (11) and (12) are shown at the bottom of next page.
ΓP = κ2 kC (7) Here, the term in the square bracket is the mechanical
impedance Zm of the coupled system. Note that the voltages
where κ2 = 1 − (fr−sc /fr−oc )2 is the electromechanical cou- generated by the PE and ED transducers are maximized when
pling factor determined from the short-circuit fr−sc and open- the system operates at its resonance (ω = ωr ), whereas the
circuit fr−oc resonant frequencies, k0 = J(2πfr−oc )2 is the impedances due to the mass moment of inertia (jωJ) and
open-circuit torsional spring stiffness, C = Cp /2 represents the torsional spring stiffness (k/jω) cancel. Using root-mean-
the free electrical capacitance of two series-connected identical square (rms) values for the torque τmag and load voltage VL ,
piezoelectric unimorphs of capacitance Cp = ∈0 wl ∈S33 /tp in corresponding time-average powers are
which ∈0 (= 8.854 × 10–12 F·m-1 ) is the permittivity of free 2
space, ∈S33 is the permittivity component at constant strain VL−PE V2
PPE = and PED = L−ED . (13)
with the plane-stress assumption of a thin beam, and w, l, and RL−PE RL−ED
tp are the width, length, and thickness of the piezo element,
The power generated by each transducer can be maximized
respectively. Due to the complicated design of the mechanical
by choosing corresponding impedance matched optimal load
suspension and displacement (rotation/translation) of the central
resistance. The value of each optimal load resistance (RL−opt )
magnet mass, J, fr−oc and fr−sc are determined via COMSOL
is equal to the magnitude of the total electrical impedance
FEA simulations. Furthermore, the torsional damping coeffi-
at the output ports of the corresponding transducer, which, at
cient b is determined from experimentally measured mechanical
resonance (ω = ωr ), are
quality factor Q as b = 2ζJωr−oc , where ζ = 1/2Q is  
the mechanical damping ratio and ωr−oc = 2πfr−oc is the  K2 
 b + RR +RRL−ED 
open-circuit angular resonant frequency. RL−PE−opt =      (14)
K 2
 Γ2P + jωr C0 b + R +RR 
R L−ED
C. Receiver Performance Analysis  
 2
KR (1 + jωr C0 RL−PE ) 
Fig. 4 presents the simplified equivalent circuit of the dual- 
RL−ED−opt = RR + 2 .
transduction receiver in order to analyze its system behavior and ΓP RL−PE + b (1 + jωr C0 RL−PE ) 
output performance, which will be compared with experimental (15)
results later for validation. In the circuit, an ideal torque source Four special cases are considered under various harmonic
is considered, which generates the torque τmag while interacts excitation and load conditions, which are explicitly described in
with the magnetic fields produced by a well-controlled source the following with the help of equivalent circuit representations
(Helmholtz coil pair) during the experiments. The complex load of the coupled system, as shown in Fig. 5.
impedances ZL−PE and ZL−ED are replaced with resistive loads
RL−PE and RL−ED for the PE and ED transducers, respectively. Case I: PE open circuit with ED open: When there is no external
The inductance LR of the receiver coil is neglected since at load connected to the ED transducer, the ED transducer does
low frequencies (<1 kHz) the coil reactance (ω LR = 4.9 Ω at not have any influence on the overall system dynamics and
744 Hz) is much smaller than the coil resistance ( RR = 71 Ω). can be omitted [see Fig. 5(a)]. Therefore, the frequency-
Finally, Zm represents mechanical impedance of the receiver, dependent no-load (open-circuit) voltage for PE transducer

Authorized licensed use limited to: College of Engineering - THIRUVANANTHAPURAM. Downloaded on August 22,2023 at 10:33:22 UTC from IEEE Xplore. Restrictions apply.
 7474 IEEE TRANSACTIONS ON POWER ELECTRONICS, VOL. 37, NO. 6, JUNE 2022

TABLE I
SYSTEM PARAMETERS USED IN THE SIMULATION

Fig. 5. Equivalent circuits of the coupled system for (a) Case I: PE open circuit
with ED open, (b) Case II: PE open circuit with ED short, (c) Case III: ED open
circuit with PE open, and (d) Case IV: ED open circuit with PE short.

properties of the components used in the dual-transduction

is reduced to EWPT receiver.
τmag ΓP
VPE−I =   . (16)
k Γ2P IV. EXPERIMENTAL VALIDATION
jωC0 b + jωJ + jω + jωC0
A. Prototype Fabrication and Assembly
Case II: PE open circuit with ED short: When the ED trans-
ducer is short-circuited, the equivalent impedance of the ED Fig. 6 shows the schematics and photographs of the dual-
2
transducer becomes ZED = KR /RR [see Fig. 5(b)]. Then, transduction EWPT receiver prototype fabrication process steps.
the frequency-dependent no-load voltage for PE transducer is The meandering suspension (1 mm wide) structure with a center
platform (2.6 × 2.6 mm2 ) and surrounding anchor was created
τmag ΓP
VPE−II =    . (17) by laser micromachining 125-μm-thick titanium (Ti) shim stock.
k Γ2P 2
KR A same-sized (2.6 × 2.6 mm2 ) 200-μm-thick silicon (Si) spacer
jωC0 b + jωJ + jω + jωC0 + RR
and a laterally magnetized N50 NdFeB magnet (5×5×1 mm3 )
Case III: ED open circuit with PE open: With the PE trans- were glued to one side of the center platform. A profession-
ducer is open-circuited, the equivalent impedance of the PE ally wound self-supported copper coil (44 AWG, 328 turns,
transducer becomes ZPE = Γ2P /jωC0 [see Fig. 5(c)]. Then, 71 Ω resistance) of inner dimension 5.6×5.6×1.4 mm3 was
the frequency-dependent no-load voltage for ED transducer glued to the anchor. On the other side of the suspension, two
becomes PZT-5A piezo-patches (5×1×0.13 mm3 ) with sputtered nickel
τmag KR electrodes, poled through the thickness (Piezo Systems, USA),
VED−III =  
Γ2P
. (18) were bonded to the clamped arms of the suspension beam
k
b + jωJ + jω + jωC 0
using electrically conductive epoxy (EO-21M-5, Epoxy Set
Inc., USA) that forms a series electrical connection between the
Case IV: ED open circuit with PE short: When the PE transducer piezo-patches. Finally, the assembled structure was bonded to a
is short-circuited, the equivalent circuit is represented using PCB, and electrical connections were created for measurements,
Fig. 5(d) and the frequency-dependent no-load voltage for ED as shown in Fig. 7.
transducer is now reduced to
τmag KR B. Experimental Setup
VED−IV =  . (19)
k
b + jωJ + jω Fig. 8 shows the photographs of the experimental setup.
The fully assembled prototype was placed at the centroid of
Table I presents the parameters used in the analytical simula- a Helmholtz coil pair (28 cm diameter, 15 cm long, 0.74
tion (lumped modeling), derived from the geometry and material mT·A-1 ). A waveform generator (Stanford Research Systems

τmag ΓP RL−PE
VL−PE =    (11)
k Γ2P RL−PE 2
KR
(1 + jωC0 RL−PE ) b + jωJ + jω + 1+jωC 0 RL−PE
+ RR +RL−ED

τmag KR RL−ED
VL−ED =    (12)
k Γ2P RL−PE 2
KR
(RR + RL−ED ) b + jωJ + jω + 1+jωC 0 RL−PE
+ RR +RL−ED

Authorized licensed use limited to: College of Engineering - THIRUVANANTHAPURAM. Downloaded on August 22,2023 at 10:33:22 UTC from IEEE Xplore. Restrictions apply.
 HALIM et al.: ANALYSIS OF A DUAL-TRANSDUCTION RECEIVER FOR ELECTRODYNAMIC WIRELESS POWER TRANSFER 7475

measured waveforms, same time scale and sampling rate were

maintained.

C. Results and Discussion

First, the open-circuit (no-load) frequency responses for the
four cases described earlier were tested at 50 μTrms magnetic
field, as shown in Fig. 9. Results indicate that the torsional
resonant frequency (fr−oc ) for the first three cases is 744.8 Hz,
whereas that (fr−sc ) for Case IV is 742.6 Hz. The ED transducer
loading conditions do not have any effect on the PE transducer
resonant frequency but affect the PE no-load voltage, depend-
ing on the equivalent ED transducer impedance. Measurement
(Meas.) results show good agreement with the analytical lumped
model simulation (Sim.). Both exhibit linear characteristics with
peaks at resonance, indicating an underdamped second-order
system with Q-factor of 90 (in air). A phase difference of ∼70°
is noted between the open-circuit voltage waveforms generated
by the PE and ED transducers, as shown in Fig. 10. This
phase difference must be taken into account while designing
an efficient downstream power management electronics, which
is out of the scope of this article.
Fig. 6. Dual-transduction receiver fabrication process steps. (a) Cross-
sectional schematics. (b) Photographs.
Next, the optimum load resistance for each transducer was de-
termined by connecting and sweeping variable load resistances
across the outputs of the corresponding transducer for the cases
described earlier. Each sweep was performed at the resonant
frequency of each respective case and under 50 μTrms field.
Fig. 11 shows how the PE load voltage and power change with
load resistance for Cases I and II, whereas Fig. 12 shows the ED
load voltage and power versus load resistance for Cases III and
IV. Measurement results show that the PE transducer yielded an
optimum load resistance of 600 kΩ (610 kΩ in simulation) for
both Cases I and II, but the maximum power generated at the
PE port was larger (8.7 μW) when the ED port was open (Case
I), and smaller (3.9 μW) when the ED port was shorted (Case
II) due to additional electromechanical damping and energy
extraction. For the ED transducer (see Fig. 12), the measured
optimum load resistance was 160 Ω for both Cases III and IV,
with similar output power values (5.4 and 5.8 μW, respectively).
Fig. 7. Photographs of a fully assembled dual-transduction receiver prototype. However, the simulation results predicted the ED optimum load
(a) Front and back sides. (b) Zoomed-in view of the top side showing electrical
connections.
resistance to be 115 Ω. This model/experimental mismatch is
attributed to experimental measurement uncertainties for the
relatively low voltages (in the range of tens of millivolt), which
propagate into larger uncertainties in the experimental power
DS345) in conjunction with a power amplifier (Crown K1) calculations.
supplied an alternating current into the Helmholtz coil pair, Once the optimum load resistances for both PE and ED
which was monitored by a current probe (Tektronix TCP312A) transducers were independently found, another set of frequency
connected to a current probe amplifier (Tektronix TCPA300). sweeps were conducted to observe the load voltage (across
The strengths of the resultant magnetic fields were measured corresponding optimum load resistances) responses of each
using a transverse Hall probe (Lakeshore XHMMA-1482) and transducer, whereas the other one was simultaneously kept open,
a Gaussmeter (Lakeshore 475DSP). An oscilloscope (Tektronix short, or optimally loaded. Fig. 13 shows the PE load voltage
DPO-2004 B) was used to measure the input current (using 1 MΩ as a function of frequency with various ED loading conditions
probe) to the Helmholtz coil pair and the output voltage (using under 50 μTrms . Results indicate that resonance occurs at the
10 MΩ probe) generated by the PE transducer of the prototype. A same frequency for all three conditions meaning that the ED
USB-driven data acquisition device (Digilent Analog Discovery transducer loading conditions do not affect the resonance of
2) was used to measure the output voltage (using 1 MΩ probe) the PE transducer, however, control the amplitude of the PE
from the ED transducer of the prototype. Simultaneously mea- load voltage (as seen earlier in Fig. 9). Similarly, Fig. 14 shows
suring outputs from both PE and ED transducers using the same the ED load voltage versus frequency with various PE loading
equipment (either oscilloscope or the data acquisition device) conditions under 50 μTrms . In this case, resonant frequency
imposes the equipment loading effect on the transducers that changes with the change in the PE loading conditions, as the
affects the actual measurement. For ease of postprocessing the model predicts. From both sets of frequency response results,

Authorized licensed use limited to: College of Engineering - THIRUVANANTHAPURAM. Downloaded on August 22,2023 at 10:33:22 UTC from IEEE Xplore. Restrictions apply.
 7476 IEEE TRANSACTIONS ON POWER ELECTRONICS, VOL. 37, NO. 6, JUNE 2022

Fig. 8. Photographs of (a) experimental setup and (b) prototype under test.

Fig. 9. No-load voltage versus frequency for various cases at 50 μTrms field. Fig. 11. PE load voltage and power versus load resistance for Cases I and II
under 50 μTrms field at 744.8 Hz resonant frequency.

Fig. 10. Measured open-circuit voltage waveforms simultaneously generated

by the PE and ED transducers under 50 μTrms field at 744.8 Hz resonance.
Fig. 12. ED load voltage and power versus load resistance for Cases III and
IV under 50 μTrms field at 744.8 and 742.6 Hz, respectively.
a new resonant frequency is obtained (743.6 Hz in simulation
and 744.6 Hz from measurement) when both transducers are
connected to their respective optimum load resistances. amplitude increases. However, while the simulation predicts a
Finally, the time-average power simultaneously delivered to quadratic response, the measured power values deviate from this
the corresponding experimentally obtained optimal load resis- trend as the field intensity increases. This behavior is attributed
tances of both PE and ED transducer were measured while to (unmodeled) nonlinear behavior of the system, including me-
varying the amplitude of the magnetic field at a fixed frequency chanical spring stiffening at higher field amplitudes, nonlinear
of 744.6 Hz. As shown in Fig. 15, both simulation and measure- piezoelectric properties for the PE transducer, and nonconstant
ment results show that the power increases as the magnetic field electrodynamic coupling (i.e., changing KR value) with change

Authorized licensed use limited to: College of Engineering - THIRUVANANTHAPURAM. Downloaded on August 22,2023 at 10:33:22 UTC from IEEE Xplore. Restrictions apply.
 HALIM et al.: ANALYSIS OF A DUAL-TRANSDUCTION RECEIVER FOR ELECTRODYNAMIC WIRELESS POWER TRANSFER 7477

Fig. 13. PE load voltage versus frequency with various ED loading conditions
under 50 μTrms field.

Fig. 16. Simulated total (PE plus ED) power as a function of both PE and ED
load resistances under 120 μTrms field at 743.6 Hz resonance.

The power transmission efficiency (PTE) of a WPT system

depends on the type of WPT technique (i.e., inductive/ ultra-
sonic, near-field/far-field, low-frequency/high-frequency) and is
heavily influenced by other factors related to the transmitter
system such as dimensions of the transmitter coil, the generated
magnetic flux density, and the distance between the transmitter
system and the receiver. A transmitter system optimized to a
receiver system for a specific WPT technique is required to
accurately determine the system PTE. In this article, we used
Fig. 14. ED load voltage versus frequency with various PE loading conditions
under 50 μTrms field. a Helmholtz coil pair to test the prototype device under very low
magnetic field intensities (≤ 120 μTrms ) for model validation,
which is not an ideal/optimized test setup to determine PTE.
However, our prior work on EWPT system with PE receiver
exhibited the maximum PTE of ∼0.036% while operating at
724 Hz frequency at 1 cm distance from the centroid of a
pancake-type transmitter coil where a 2 mTrms magnetic field
(within safe human exposure limit) was generated [15]. We
admit that the PTE of the EWPT system is low compared to that
of the conventional inductive/ultrasonic WPT techniques [20];
however, an EWPT system can potentially overcome the draw-
backs/limitations of conventional WPT systems, especially for
wearable and implantable medical devices (e.g., cardiac pace-
maker, deep brain neurostimulator, cochlear implants, etc.) ap-
plications. This low-frequency EWPT method facilitates higher
Fig. 15. PE and ED load power versus magnetic field simultaneously delivered
magnetic field safety margins tightly limited by standard regu-
to corresponding optimal load resistances at resonance (743.6 Hz for simulation lations [8]–[10], and conceals the field attenuation and parasitic
and 744.6 Hz for measurement). heating effects while transmitting through conductive media,
e.g., metal enclosure, human body, etc. [15].
Note that the optimum load resistances of the PE and ED
transducers were determined independently and used in the
of field amplitudes for the ED transducer. Nevertheless, the analytical simulation to obtain the simulated power values from
maximum experimentally measured powers are 35 and 14 μW the corresponding transducer (solid lines in Fig. 15). However,
from the PE and ED transducers, respectively (combined total for this coupled, two-output-port system, an important question
of 49 μW) at 120 μTrms field. According to the power versus is raised: Are the independently determined optimum load re-
magnetic field plots presented in Fig. 15, the power level can sistances truly the optimum? In other words, is the total system
be further scaled up by increasing the interacting magnetic output power (i.e., PE power + ED power) actually maximized?
field amplitudes transmitted from the transmitter system, as Using the lumped model, we further extended our analysis to
reported in [19]. Additionally, the power can be increased by predict the total maximum power by simultaneously varying
increasing the mechanical Q-factor of the receiver, which is done both load resistances. Fig. 16 shows the simulated total power
by optimizing the suspension design. (PE power + ED power) while simultaneously varying the PE

Authorized licensed use limited to: College of Engineering - THIRUVANANTHAPURAM. Downloaded on August 22,2023 at 10:33:22 UTC from IEEE Xplore. Restrictions apply.
 7478 IEEE TRANSACTIONS ON POWER ELECTRONICS, VOL. 37, NO. 6, JUNE 2022

TABLE II
COMPARISON WITH PREVIOUSLY REPORTED EWPT RECEIVERS

and ED load resistances at 743.6 Hz resonant frequency and 120 REFERENCES

μTrms magnetic field. Results show that the simulated maximum
[1] J. Siang, M. H. Lim, and M. S. Leong, “Review of vibration-based energy
total power of 65 μW is obtained when the PE and ED load harvesting technology: Mechanism and architectural approach,” Int. J.
resistances are tuned at 580 kΩ and 230 Ω, respectively (red dot Energy Res., vol. 42, no. 5, pp. 1866–1893, Apr. 2018.
on the surface plot), whereas with the independently simulated [2] T. Ghomian and S. Mehraeen, “Survey of energy scavenging for wearable
optimum load resistances (600 kΩ for PE and 160 Ω for ED), the and implantable devices,” Energy, vol. 178, pp. 33–49, Jul. 2019.
[3] J. Huang, Y. Zhou, Z. Ning, and H. Gharavi, “Wireless power transfer and
total power is 63 μW (black dot on the surface plot). This result energy harvesting: Current status and future prospects,” IEEE Wireless
is important because it shows that truly optimizing the system Commun., vol. 26, no. 4, pp. 163–169, Aug. 2019.
power is not as simple as just optimizing one port and then the [4] M. Song, P. Belov, and P. Kapitanova, “Wireless power transfer inspired
other port. Specifically, for the system studied here, the “true” by the modern trends in electromagnetics,” Appl. Phys. Rev., vol. 4, no. 2,
optimum ED load resistance is 44% larger, and the PE load Jun. 2017, Art. no. 021102.
[5] A. M. Jawad, R. Nordin, S. K. Gharghan, H. M. Jawad, and M. Ismail,
resistance is 3.4% smaller than what would be expected from “Opportunities and challenges for near-field wireless power transfer: A
analyzing each port individually. It is anticipated that this effect review,” Energies, vol. 10, no. 7, Jul. 2017, Art. no. 1022.
is strongly correlated with the strength of electromechanical [6] D. Kim, A. Abu-Siada, and A. Sutinjo, “State-of-the-art literature review
coupling of the transducers, and should be carefully considered of WPT: Current limitations and solutions on IPT,” Elect. Power Syst. Res.,
vol. 154, pp. 493–502, Jan. 2018.
in future designs. [7] N. Garraud, A. Garraud, D. Munzer, M. Althar, and D. P. Arnold,
“Modeling and experimental analysis of rotating magnet receivers for
electrodynamic wireless power transmission,” J. Phys. D, Appl. Phys.,
vol. 52, no. 18, Feb. 2019, Art. no. 185501.
V. CONCLUSION [8] IEEE Standard for Safety Levels With Respect to Human Exposure to
Electric, Magnetic, and Electromagnetic Fields, 0 Hz to 300 GHz, IEEE
In this article, we have demonstrated a dual-transduction Standard C95.1-2019, 2019.
EWPT receiver for low-frequency, near-field WPT systems [9] ICNIRP, “Guidelines for limiting exposure to time-varying electric and
by developing an electromechanical LEM, which is then vali- magnetic fields (1 Hz to 100 kHz),” Health Phys., vol. 99, no. 6,
dated by experimental characterization of a fabricated prototype. pp. 818–836, Dec. 2010.
[10] ICNIRP, “Guidelines for limiting exposure to electromagnetic fields,”
This dual-transduction receiver simultaneously generates volt- Health Phys., vol. 118, no. 5, pp. 483–524, May 2020.
age/power from a PE transducer and an ED transducer while [11] B. D. Truong, D. Wang, T. Xue, S. Trolier-McKinstry, and S. Roundy,
operating at its torsional vibration mode by the influence of “Experiments on a wireless power transfer system for wearable device
an external time-varying magnetic field. From the analyses with sol-gel thin-film PZT,” J. Phys., Conf. Ser., vol. 1407, Dec. 2019,
as well as the experimental characterization, it is observed Art. no. 012063.
[12] M. A. Halim, S. E. Smith, J. M. Samman, and D. P. Arnold, “A high-
that the maximum transferable power under a given magnetic performance electrodynamic micro-receiver for low-frequency wireless
field can be achieved by concurrently tuning the driving fre- power transfer,” in Proc. IEEE Int. Conf. Micro Electro Mech. Syst.,
quency and adapting the load resistance for each transducer. Vancouver, BC, Canada, 2020, pp. 590–593.
Compared with prior works, as summarized in Table II, this [13] H. A. C. Tilmans, “Equivalent circuit representation of electromechanical
transducers: I—Lumped-parameter systems,” J. Micromech. Microeng.,
article successfully blended the two transduction schemes (PE vol. 6, no. 1, pp. 157–176, Mar. 1996.
and ED) within a low-profile, volume-efficient architecture and [14] S. D. Senturia, “Part II modeling strategies,” in Microsystem Design.
maximizes the normalized power density (NPD) for wirelessly Dordrecht, The Netherlands: Kluwer, 2001, pp. 103–178.
charging biomedical implants. The device generates maximum [15] M. A. Halim, A. A. Rendon-Hernandez, S. E. Smith, and D. P. Arnold,
49 μW power at its torsional resonance of 744.6 Hz while op- “A chip-sized piezoelectric receiver for low-frequency, near-field wireless
power transfer: Design, modeling and experimental validation,” Smart
erating under 120 μTrms field, which is well below the standard Mater. Struct., vol. 30, no. 4, Mar. 2021, Art. no. 045011.
human exposure limit (∼2.8 mTrms at 744.6 Hz). This power [16] B. D. Truong and S. Roundy, “Wireless power transfer system with center-
generation corresponds to a power density of 0.54 mW·cm–3 clamped magneto-mechano-electric (MME) receiver: Model validation
and an NPD of 37 mW·cm-3 ·mT–2 . Experimental results show and efficiency investigation,” Smart Mater. Struct., vol. 28, no. 1, Jan. 2019,
Art. no. 015004.
good agreement with the model predictions. However, some [17] M. A. Halim, A. A. Rendon-Hernandez, S. E. Smith, J. M. Sam-
deviations have been observed (22% maximum), especially man, N. Garraud, and D. P. Arnold, “Miniature electrodynamic wire-
when the magnetic field is increased, which is attributed to me- less power transmission receiver using a micromachined silicon sus-
chanical and electromechanical nonlinearities that require future pension,” J. Microelectromech. Syst., vol. 30, no. 1, pp. 144–155, Feb.
analysis. 2021.

Authorized licensed use limited to: College of Engineering - THIRUVANANTHAPURAM. Downloaded on August 22,2023 at 10:33:22 UTC from IEEE Xplore. Restrictions apply.
 HALIM et al.: ANALYSIS OF A DUAL-TRANSDUCTION RECEIVER FOR ELECTRODYNAMIC WIRELESS POWER TRANSFER 7479

[18] M. A. Halim, A. A. Rendon-Hernandez, and D. P. Arnold, “An electro- Spencer E. Smith was born in Fort Benning, GA,
dynamic wireless power receiver ‘chip’ for wearables and bio-implants,” USA, in 1998. He received the dual B.S. degrees in
in Proc. IEEE PELS Workshop Emerg. Technol., Wireless Power Transf., mechanical and aerospace engineering in 2020 and
Seoul, South Korea, 2020, pp. 271–274. the M.S degree in aerospace engineering in 2022 with
[19] S. E. Smith, M. A. Halim, A. A. Rendon-Hernandez, and D. P. Arnold, a focus on dynamics, systems, and control from the
“Dual-transduction electromechanical receiver for near-field wireless University of Florida, Gainesville, FL, USA.
power transmission,” in Proc. IEEE Int. Conf. Micro Electro Mech. Syst., Since 2019, he has been with the Interdisciplinary
2021, pp. 38–41. Microsystems Group, University of Florida, with
[20] R. V. Taalla, M. S. Arefin, A. Kaynak, and A. Z. Kouzani, “A review a focus on the design, fabrication, and testing of
on miniaturized ultrasonic wireless power transfer to implantable medical wireless power receivers and hybrid piezo/magnetic
devices,” IEEE Access, vol. 7, pp. 2092–2106, 2018. electromechanical transformers. As a member of the
[21] N. Garraud, D. Alabi, J. D. Varela, D. P. Arnold, and A. Garraud, “Micro- National Science Foundation’s Internet of Things for Precision Agriculture
fabricated electrodynamic wireless power receiver for bio-implants and (IoT4Ag). Since 2020, he also worked on electrodynamic wireless power
wearables,” in Proc. Solid-State, Actuators, Microsyst. Workshop Tech. transmission technologies for agricultural sensors and aerial drones.
Dig., Hilton Head Island, SC, USA, May 2018, pp. 34–37.
[22] M. A. Halim, J. M. Samman, S. E. Smith, and D. P. Arnold, “Piezoceramic
electrodynamic wireless power receiver using torsion mode meandering
suspension,” in Proc. Int. Conf. Micro Nanotechnol. Power Gener. Energy
Convers. Appl., Krakow, Poland, Dec. 2019, pp. 1–5, doi: 10.1109/Pow-
erMEMS49317.2019.20515809768.
[23] G. Liu, P. Ci, and S. Dong, “Energy harvesting from ambient low frequency
magnetic field using magneto-mechano-electric composite cantilever,” David P. Arnold (Senior Member, IEEE) received
Appl. Phys. Lett., vol. 104, no. 3, Jan. 2014, Art. no. 032908. the dual B.S. degrees in 1999, the M.S. degree in
[24] S. Du, E. K. Chan, B. Wen, J. Hong, H. Widmer, and C. E. Wheatley, 2001, from the University of Florida, Gainesville, FL,
“Wireless power transfer using oscillating magnets,” IEEE Trans. Ind. USA, and the Ph.D. degree from the Georgia Institute
Electron., vol. 65, no. 8, pp. 6259–6269, Aug. 2018. of Technology, Atlanta, GA, USA, in 2004, all in
electrical and computer engineering.
He is currently the George Kirkland Engineering
Miah A. Halim received the B.S. and M.S. degrees Leadership Professor and Associate Chair for Re-
in electrical and electronic engineering from Islamic search with the Department of Electrical and Com-
University, Kushtia, Bangladesh, in 2008 and 2010, puter Engineering, University of Florida, the Direc-
respectively, and the Ph.D. degree in electronic engi- tor of University/Industry Partnerships and UF Site
neering in 2016 from Kwangwoon University, Seoul, Director, NSF IoT4Ag ERC, and a member of the UF Interdisciplinary Mi-
South Korea. crosystems Group. He coauthored more than 200 refereed journal and conference
Later, he worked on energy harvesting for wearable publications, and holds more than 20 U.S. patents. His research interests include
smart devices applications with Mechanical Engi- micro/nanostructured magnetic materials, magnetic microsystems, electrome-
neering Department, University of Utah, Salt Lake chanical transducers, and miniaturized power/energy systems.
City, UT, USA. He is currently a Postdoctoral Re- Dr. Arnold’s research innovations have been recognized by the 2008 Presi-
search Associate and a member of the Interdisci- dential Early Career Award in Science and Engineering and the 2009 DARPA
plinary Microsystems Group, Electrical and Computer Engineering Department, Young Faculty Award. He is an active participant in the magnetics and MEMS
University of Florida, Gainesville, FL, USA. His current research interests communities, with ongoing involvement in various conference committees and
include magnetic microsystems for low-power applications. His main research currently serving on the editorial boards of Micromachines and Journal of
interests include electromechanical transducers, multisource energy harvesting, Micromechanics and Microengineering. He is also a member of Tau Beta Pi
wireless power transfer systems, and multifunctional materials and systems. and Eta Kappa Nu.

Adrian A. Rendon-Hernandez received the B.S.

degree in electronics and communication engineering
(with the highest distinction) from the Autonomous
University of Nuevo Leon, San Nicolas de los Garza,
Mexico, in 2008, the M.S. degree in electronics sys-
tems engineering from the Monterrey Institute of
Technology, Monterrey, Mexico, in 2012, and the
Ph.D. degree in nanoelectronics and nanotechnolo-
gies (with jury congratulations) in 2018 from the
Universit Grenoble Alpes, Grenoble, France.
During his career, he held several engineering po-
sitions with Schneider Electric and AT&T-Alestra. He was also a Teaching
Assistant with Polytech Grenoble and the Monterrey Institute of Technology. He
is currently a Postdoctoral Research Associate and a member of the Interdisci-
plinary Microsystems Group, Electrical and Computer Engineering Department,
University of Florida, Gainesville, FL, USA. His research interests include
design, modeling, multiphysics simulation, and characterization of wireless
power transmission and energy harvesting devices.

Authorized licensed use limited to: College of Engineering - THIRUVANANTHAPURAM. Downloaded on August 22,2023 at 10:33:22 UTC from IEEE Xplore. Restrictions apply.