Xiaoqi Bao1, Will Biederman, Stewart Sherrit, Mircea Badescu, Yoseph Bar-Cohen, Christopher Jones, Jack Aldrich and

Zensheu Chang Jet Propulsion Laboratory, California Institute of Technology, 4800 Oak Grove Drive, Pasadena, CA 91109, USA
ABSTRACT
Piezoelectric acoustic-electric power feed-through devices transfer electric power wirelessly through a solid wall using elastic waves. This approach allows for the elimination of the need for holes through structures for cabling or electrical feed-thrus . The technology supplies power to electric equipment inside sealed containers, vacuum or pressure vessels, etc where holes in the wall are prohibitive or may result in significant performance degradation or requires complex designs. In the our previous work, 100-W of electric power was transferred through a metal wall by a small, piezoelectric device with a simple-structure. To meet requirements of higher power applications, the feasibility to transfer kilowatts level power was investigated. Pre-stressed longitudinal piezoelectric feed-thru devices were analyzed by finite element modeling. An equivalent circuit model was developed to predict the characteristics of power transfer to different electric loads. Based on the analytical results, a prototype device was designed, fabricated and successfully demonstrated to transfer electric power at a level of 1-kW. Methods of minimizing plate wave excitation on the wall were also analyzed. Both model analysis and experimental results are presented in detail in this paper. KEYWORD: piezoelectric devices, acoustic wave, electric power supply, wireless power feed, pressure vessels.

High-power piezoelectric acoustic-electric power feedthru for metal walls

1. INTRODUCTION
There are many engineering applications where the use of wires to transfer power and communicate data through the walls of a structure is prohibitive or involves a significantly complex design. The use of feed-through wires in such systems may make them prone to leakage of chemicals or gasses, loss of pressure or vacuum, as well as hamper the ability to adequately insulate them thermally or electrically. Various future NASA missions are expected to require transmission of power into sealed solid metallic structures. Such structures may include a sample container providing planetary protection that requires internal power for monitoring, or power a spacecraft or power the space station in rendezvous and docking, as well as support autonomous operations. To address this need the method of acousticelectric transmission using elastic waves to transmit power was investigated. This transmission device uses the direct and indirect piezoelectric effects as means of generating stress waves that are transmitted through walls where the received wave is converted to an electric power using a piezoelectric transducer and is delivered to an electric load. Potentially, the technology will allow for both power and/or data transfer from either direction. In previous work we demonstrated the transmission of 100 W, at 87-88% efficiency through a 3.4 mm thick titanium plate using two piezoelectric ceramic disks [Bao et. Al. 2007]. In this paper, a 1kW feed though by pre-stressed PZT transducer with 88% efficiency is reported.

2. BACKGROUND
Using elastic waves to transfer power through a wall was suggested [Hu et al, 2003]. A system having piezoelectric layers on both sides of a plate, one as a transmitter and the other as a receiver, was investigated. A sinusoidal voltage was applied across the transmitting piezoelectric layer generating an elastic wave that travels through the wall into the receiving piezoelectric transducer where the stress wave generates a sinusoidal voltage. The theoretical problem of a piezoelectric/elastic layer/piezoelectric was solved using 1-D wave equations. An alternative approach based on network equivalent circuits was suggested [Sherrit et al. 2005, 1999] that can be easily modified to account for

Correspondence: xbao@jpl.nasa.gov
Industrial and Commercial Applications of Smart Structures Technologies 2008, edited by L. Porter Davis, Benjamin Kyle Henderson, M. Brett McMickell, Proc. of SPIE Vol. 6930, 69300Z, (2008) 0277-786X/08/$18 doi: 10.1117/12.776473 Proc. of SPIE Vol. 6930 69300Z-1

2008 SPIE Digital Library -- Subscriber Archive Copy

additional system elements such as backing and connected directly to other networks or circuits. The possible loss mechanisms of the solution can be accounted for and introduced into the model. In previous work we demonstrated the transmission of 100 W, through a 3.4 mm thick titanium plate using two ceramic disk piezoelectric transducers [Bao et al 2007]. Several wireless piezoelectric acoustic-electric power feedthru devices were fabricated and tested. Test results showed that 100-W power feed-though is feasible for the device using two 38-mm diameter PZT disks. To demonstrate the feasibility of powering a realistic load, a 100W light bulb was lit using the power transferred by the unit (see Figure 1). This demonstration was executed by applying an electric voltage to the transmitting piezoelectric disk at 747 kHz.

No power to piezoelectric transmitter. Figure 1. 100-Watt light bulb demonstration.

Piezoelectric transmitter excited

3. ANALYSIS OF HIGH POWER FEEDTHRU DEVICE

In order to increase the power transmission to the level of a kW, it is necessary to enhance and optimize the original device. For the previous tested device the level of stress necessary to produce power at the 1 kW level will exceed the strength of the piezoelectric material and the bond joints between the piezoelectric element and the wall. In order to counter this effect, pre-stressed longitudinal piezoelectric transducers were selected. As shown in Figure 2, the transducer consists of a piezoelectric ceramic stack, front and back mass blocks and a bolt. The stack is compressed by the bolt. Since the compressive strength of ceramics is much larger than the tensile strength, the elements pre-stressed under compression will withstand much higher level of alternative stress changes excited by input voltage in energy transmission operation. The diameter, thickness and the number of ceramic elements of the stack can be adjusted to meet the power requirement and obtain proper impedance to match the source and the load.

Mass block Bolt

Mass block

Piezo stack Wall Figure 2. Sketch of feed thru device consisting of two pre-stressed longitudinal transducers The vibration characteristics of the design were analyzed by finite element model. ANSYS MutiPhysics software was used. The feedthru device was modeled as 2-D axisymmetric model. To determine the design parameters, the model

Proc. of SPIE Vol. 6930 69300Z-2

included all the parts of the transducers and a small part of the wall was modeled in modal analysis. The operating frequency was selected from the resonance modes according to the frequency, effective electromechanical coupling factor and impedance for best transmission performance. Figure 3 shows the mode shape of the selected resonance of the determined design, where an asymmetric mode was used (in reference to the wall) and in the figure only half of the device is shown.

. Figure 3. Half of the device is shown with the shape of the 24.7 kHz mode. This is an asymmetric mode to the wall. In addition, an equivalent circuit model was developed to predict the energy transfer performance to various loads (Figure 4). The circuit represents the characteristics of the feedthru device at the frequency near its resonance. The parameters of the circuit were determined by the finite element analysis or the parameter fitting function of impedance analyzer such as Agilent 4294A. In the case where the transducers on two sides of the wall are identical, the C0 = C2. The C0 and C2 could include dielectric loss of the piezoelectric material. It should be pointed out that the R1 represents the loss cause by the vibration and should include the material loss of the device and the energy loss due to the associated waves on the wall plate. The plate wave loss was analyzed by a finite element model including a larger wall as detailed in section 5. The material loss internal to the device is nonlinear and, usually, highly dependent on the amplitude of vibration, which should be measured experimentally for accuracy.

L1

C1

R1

V ~

C0

C2

RL

Figure 4. Equivalent circuit for the feedthru device near resonance frequency

Proc. of SPIE Vol. 6930 69300Z-3

4. TEST SAMPLES AND EXPERIMENTAL RESULTS

Based on the analysis results, an experimental device of wireless feedthru was made consisting of two sets of piezoelectric stacks that are 50 mm in diameter. As the device is symmetric, either stack may serve as the power transmitter or the receiver. The simulated wall consists of a 5 mm thick titanium plate 85 mm in diameter with a 20 mm thick 55 mm diameter sections on either side of the plate to thread the stress bolts and mechanically attach the piezoelectric ceramic rings. In order to reduce the fabrication costs for this prototype, the diameter of the titanium plate was kept to a diameter of 85 mm. This small diameter also means that the radial surface waves in the plate generated from the vibrations are not dissipated and are instead reflected when they reach the edge of the plate. Therefore, this prototype we will not suffer from significant plate wave loss, which we will discuss later In order to prove the feasibility of transmitting 1 kW with this method for transmitting power acoustically, a test setup (shown in Figure 5) was designed and built to measure the power transmissions and the power transfer efficiency. The input waveform was generated from a signal generator feeding two ENI 1140L amplifiers in order to generate enough input power. The two amplifiers were connected as shown in the figure. By connecting the grounds together and feeding 180 out of phase input signals, the voltage between the two outputs of the amplifiers was doubled and applied to the electrodes of the transmitter transducer. On the receiving end, a load resistor is connected across the output electrodes to simulate the resistance of the device we are tying to transmit power to. To measure the input and output power, a four channel digital oscilloscope Tektronix TDS5054B is used. The input and output voltages are measured using high voltage probes, and the input current is measured using a current probe. With the two amplifier configuration, the electric common of the feedthru device was not grounded. The output voltage was calculated as the voltage difference of the two electrodes of the receiving transducer. Using the measured data, we can calculate the input and output power of the device, and the transmission efficiency. The formulas used to calculate the input power, output power, and transfer efficiency are presented in Figure 5. The integrating time T is an integer number of the cyclic period of the signals. The calculations were performed by the oscilloscope.
Signal Generator CH1 Input Amplifier #1
Scope CH 1 Input voltage 1 Scope CH 3 Input Current Scope CH 4 Output voltage + Input voltage2

Rload

Input Amplifier #2 Signal Generator CH2 (180 out of phase from CH1)
Scope CH 2 Input voltage 2
T 2

Efficiency =

Pout = Pin

V Rout dt / T load 0 VCH 2 ) I in dt / T

(V
0

(CH 4 CH 2) 2 ]/T Rload = Area[(CH1 CH 2) I in ] / T Area[

CH 1

Figure 5. Test bench setup and power calculations Preliminary tests found that our measurement setup was yielding overestimates in the efficiency calculation since the method we used to calculate efficiency is highly dependent on the phase of the waveforms. It was determined that the current probe was causing phase errors for our input power calculations. A MatLab script was written to perform a

Proc. of SPIE Vol. 6930 69300Z-4

correction of the results calculated by the oscilloscope. In latter tests we used a new high power metal film resistor with extremely low inductance to sample the current, which was found to reduce the error. In order to automate these measurements a LabVIEW program was written to control the instruments. The system is able to scan frequency and find optimal operating frequency. The most efficiency operating frequency was found at 24.5 kHz. At this frequency the test device transferred 1083 W to a load resistor of 300 . The efficiency was calculated as 84% after correction. The screen of the oscilloscope recorded in the test is shown in Figure 6. There are the traces of the four channels and variables created in the input and output power calculations. The peaks of input voltages Ch1 and Ch2 are over the screen range but the inner data was not truncated. The waveforms were greatly distorted because the amplifiers were over driven to obtain maximum output power. The area(M2) 1083kvvH on the screen indicates the calculated output power of 1083 W. It is believed that the efficiency should be increased over the value we measured if harmonics in the driving voltage could be reduced.

I UWACq trig .iw . l'.n ii.

a.. s

Figure 6. The waveforms of power feed thru test to 300 load at 24.5 kHz, 1083 Watts at 84% efficiency. To demonstrate visually the devices ability to supply power to a realistic load, ten 120V 100W light bulbs were lit using the test device. The bulbs were divided into two sets. Each set had five bulbs serially connected. The two set were connected to the output of the test device in parallel. This demonstration was executed by applying a voltage to the transmission device at 24.5 kHz. A photograph of the excited light bulbs is shown in Figure 7. The output voltage on the load of the light bulbs was measured as 580 V rms.

Figure 7. Ten 100W light bulbs powered by the wireless transmission device.

Proc. of SPIE Vol. 6930 69300Z-5

5.

PLATE WAVE LOSS AND COMPENSATION

The tested device has a small flange to represent the wall. In feasible structures acoustic radiation in the wall may cause loss of energy and a decrease in efficiency of the device. Vibration of the PZT transducer on the wall of the metallic structures works as a source of elastic waves. The elastic waves for the case under consideration are plate waves called Lamb waves. The propagating modes of the Lamb wave are capable of traveling along plate-like structure and carrying energy away from the source. If the dimension of the metallic structure is much larger than the wavelength, then most of the energy carried away by the Lamb waves will be lost in the structure. In a previous paper [Chang et al 2007], the wave radiation energy due to a stress in a area of plate was estimated. That result corresponds to the case where the feedthru device is excited in a symmetric resonance mode with respect to the plate. In the case discussed in previous section, the tested prototype with longitudinal transducers was operated in an asymmetric mode. To predict losses for the asymmetric mode, we remodeled the device with a large wall structure and looked at the effect of using a step ring reflector in the plate to reduce the energy losses. The FEM model is shown in Figure 8. In the FE model, the transmitted wall was assumed to be very large compared to the diameter of the feedthru device and the wall was modeled as a circular plate with a diameter 30 times larger than the diameter of the PZT transducer. The material of the wall was assumed to be titanium. Further, the plate was divided into 9 rings with gradually increasing damping coefficients as the radius increases. This arrangement was used to allow the elastic wave energy to be absorbed and the reflections from the edge will be negligible. A harmonic analysis was performed for the case of the feedthru device on 4.76-mm (3/16) thick titanium wall. A pair of steel rings was included in the model. Figure 9 shows the plate wave propagating on the wall with no reflector rings. A parametric study was performed with various ring thicknesses. As presented in Figure 10, the results show the rings can effectively reduce the loss caused the plate waves. The energy loss was 25% with no ring reflector and can be reduced to 0.2% by 18-mm thick rings with 100mm inner diameter and 20mm width.

Figure 8. The FE model including the prototype piezoelectric device, 4.76-mm Titanium wall, reflector rings and a load resistor of 300 .

Proc. of SPIE Vol. 6930 69300Z-6

Figure 9. The plate wave propagates from the device area and transmits energy into the structure.

Thick nes s (m ) 0 1 0.002 0.004 0.006 0.008 0.01 0.012 0.014 0.016 0.018 0.02

0.1

Loss 0.01 0.001

Figure 10. Loss due to plate waves vs. thickness of steel rings with 100mm inner diameter and 20mm wide on 3/16 titanium wall

Proc. of SPIE Vol. 6930 69300Z-7

6. SUMMARY
This study focused on developing the capability to transmit KW levels of electric power through metal walls using elastic waves. Our results show that transmitting power at levels of more than 1kW with reasonable high efficiency is feasible using a device that consists of two sets of 50 mm diameter pre-stressed piezoelectric transducer. The issue of energy loss caused by plate waves was investigated by finite element model. The results show the loss can be quite large. However, reflectors can effectively reduce the plate wave energy loss. In future work the function of reflectors to minimize loss of the plate waves will be evaluated experimentally. In addition thermal analysis, stability studies, life test and further optimization of design are required for the successful application of the device.

ACKNOWLEDGMENT
Research reported in this manuscript was conducted at the Jet Propulsion Laboratory (JPL), California Institute of Technology, under a contract with National Aeronautics Space Agency (NASA). The authors would like to express their appreciation of Neville Marzwell, Program Manager, Advanced Concepts-Innovative Technology Programs (ESMD) for supporting this reported task

REFERENCES
Bao X., B. J. Doty, S. Sherrit, M. Badescu, Y. Bar-Cohen, J. Aldrich, and Z. Chang Wireless piezoelectric acousticelectric power feedthru Proc. SPIE 6529, 6529-40 (2007) Chang Z., X. Bao, B. J. Doty, S. Sherrit, Y. Bar-Cohen, M. Badescu, and J. Aldrich Power loss considerations in wireless piezoelectric acoustic-electric power feedthru, SPIE Smart Structures and Materials & Nondestructive Evaluation and Health Monitoring 14th International Symposium, Proceedings of SPIE - The International Society for Optical Engineering, v 6529-136, 2007 Hu Y., X. Zhang, J. Yang, Q. Jiang, Transmitting Electric Energy Through a Metal Wall by Acoustic Waves Using Piezoelectric Transducers, IEEE Transactions on Ultrasonics, Ferroelectrics, and Frequency Control, 50, 7, pp. 773781, 2003. Sherrit S., S. P. Leary, B. P. Dolgin and Y. Bar-Cohen, Comparison of the Mason and KLM Equivalent Circuits for Piezoelectric Resonators in the Thickness Mode, Proceedings of the IEEE Ultrasonics Symposium, Lake Tahoe, 1999, pp. 921-926. Sherrit, S., Badescu, M.; Xiaoqi Bao; Bar-Cohen, Y.; Chang, Z., Efficient electromechanical network model for wireless acoustic-electric feed-throughs, Proceedings of the SPIE - The International Society for Optical Engineering, v 5758, n 1, 2005, p 362-372

Proc. of SPIE Vol. 6930 69300Z-8