Design and evaluation of linear ultrasonic motors for a cardiac compression assist device

Sensors and Actuators A 119 (2005) 214 220 Design and evaluation of linear ultrasonic motors for a cardiac compression assist device Yang Ming a,, Zhu Meiling a, Robert C. Richardson a, Martin C. Levesley a, Peter G. Walker a, Kevin Watterson b a School of Mechanical Engineering, University of Leeds, Leeds LS2 9JT, UK b Yorkshire Heart Centre, Leeds General Infirmary, Leeds, UK Received 7 February 2003; received in revised form 26 August 2004; accepted 7 September 2004 Abstract Exploiting new suitable actuators is key in developing cardiac compression assist devices. In this paper, a self-moving linear ultrasonic motor with large displacement within a limited housing space is proposed and investigated as a primary attempt to develop a new actuator suitable for a cardiac compression assist device. In the motor, the rectangular piezoelectric ceramic plates are directly used as the vibrators of a self-moving element with just one input power. The housing is used directly as a guide. Reversible motion direction is achieved through applying power to different electrodes. The motion principle is analysed on the basis of harmonic response calculation results. Using a data acquisition board with LabVIEW software, a control circuit, a laser displacement sensor, a force sensor, and a spring, an experimental rig is set up. Measured resonant frequencies and frequency properties agree with the calculated results and their analysis. Also under force control of the linear ultrasonic motor, 1 Hz pulse of desired force is realised with thrust force changing from 0 to 68% of the desired force in about 100 ms after 8 mm of displacement. 2004 Elsevier B.V. All rights reserved. Keywords: Linear ultrasonic motors; Cardiac compression assist devices; Piezoelectric ceramics 1. Introduction Heart disease is a major health problem in the world. Although heart transplantation is an accepted method to treat severe cases of the disease, the demand for heart transplants exceeds the supply. Therefore, some other devices are required to assist the diseased heart [1,2]. Although a variety of ventricular assist devices are available now, most of these devices require direct contact with patient s blood; thus, thromboembolic events, anticoagulationand hemolysis-related complications, and immune reactions are ever present problems. Accordingly, there has been renewed interest in the development of techniques to support the circulation by compressing the weakened heart from its Corresponding author. Tel.: +44 113 233 2206; fax: +44 113 242 4611. E-mail address: menmy@leeds.ac.uk (Y. Ming). epicardial surface or direct cardiac compression [2]. One attempt [3] has been made as a cardiac compression assist device, where skeletal muscle is wrapped around the diseased heart and contracts to squeeze the diseased heart. This technique was shown to be partly successful in that the heart was briefly assisted. Unfortunately, skeletal muscle fatigues rapidly and so mechanical alternatives have been developed, such as the pneumatic Direct Cardiac Compression device and the Electro-Active Polymer device [4,5]. However, the pneumatic device is bulky and prone to leaking, and the force generated by Electro-Active Polymer is too small at moment. To improve the quality of life, it is necessary to exploit new actuators suitable for such an application. Ultrasonic motors have many unique characteristics, such as high power density, fast response, easy miniaturisation, and silent motion [6 15], which make them potentially very attractive for being an actuator for the cardiac compression assist device. Since the 0924-4247/$ see front matter 2004 Elsevier B.V. All rights reserved. doi:10.1016/j.sna.2004.09.005

Y. Ming et al. / Sensors and Actuators A 119 (2005) 214 220 215 application of ultrasonic motors in cardiac compression assist devices is still in its early stage, very few related publications have been found to date. After many years of investigation, ultrasonic motors are already commercially available, including applications in autofocus of cameras and precise positioning [7]. There are many types of ultrasonic motors [8 10] such as travelling wave, standing wave, hybrid transducer, and multi-vibration mode. Generally, all of their vibrators are metal and piezoelectric ceramic composites, in which elliptical motion is generated by the superposition of the two resonant vibrations excited by two ultrasonic transducers or two groups of piezoelectric elements. Hence, these motors maybe too heavy for implantation as cardiac compression assist devices. Also they usually have two power inputs with out of phase for the motion direction control. However, this approach increases the complexity of the structure and power amplifier. Further, without special design, their shapes do not usually fit the space left in the cavity around the heart, making them unsuitable as an actuator for a cardiac compression device. Therefore, the ultrasonic motor must be specially designed to fit the requirements of the cardiac compression assist devices. In this paper, a self-moving linear ultrasonic motor with large displacement housed in a guide is proposed and investigated as a primary attempt to make a suitable actuator of the cardiac compression assist device. In Section 2, the design requirements of linear ultrasonic motor are discussed firstly from the view of cardiac compression assist devices. Using calculation results of finite element software, ANSYS, Section 3 describes design and configuration of a self-moving rectangular plate linear ultrasonic motor and its motion analysis. Section 4 presents the experimental rig, composing of computer, data acquisition board using LabVIEW software, a control circuit, laser displacement sensor and force sensor. Experimental results are also discussed in Section 4. 2. Design considerations The first step in the design of the ultrasonic motor is to understand the requirements from the view of heart assist [11]. The average heart rate is about 60 75 beats per min with one beat lasting around 1 0.8 s. The cardiac cycle is divided into two parts, systole lasting around 1/3 and diastole lasting around 2/3. Blood ejection from the heart occurs during systole, and hence the rate of ventricular volume change is approximately a skewed half sine wave lasting around 300 ms. If a heart cannot pump sufficient blood to sustain life, a cardiac compression assist device wrapping around the diseased heart can provide some assistance. A linear ultrasonic motor with enough thrust force and fast response could be used as an actuator for cardiac compression assist devices. Another requirement is that the size, weight and displacement of the device must be compatible with its implantation position. Generally, there are three approaches to construct linear ultrasonic motor [12 18]. In the first one, travelling wave mo- Fig. 1. Configuration of a direct cardiac compression device element based on the linear ultrasonic motors. tors use two power inputs with π/2-phase difference, which has been extensively applied in rotary ultrasonic motor. However, the travelling wave approach has rarely been used for a linear ultrasonic motor due to its complicated mechanical design and low efficiency. Standing wave motor generates driving force from the elliptic vibration of surface points and a normal force. It usually generates larger thrust force and its electrical excitation may be very simple, however the mechanical structure may be complex with a non-reversible motion direction. The third approach uses combination vibration modes to produce elliptical motion such as longitudinal and bending vibration. Usually, in this approach, a carefully designed rectangular plate vibrator is used to superpose two resonant vibrations to produce the elliptical motion. Therefore, to make motors suitable for implantation, a thin and lightweight rectangular plate linear ultrasonic motor is chosen as the actuator for the cardiac compression assist device. The heart muscle contracts and relaxes in beats. To assist the diseased heart, it is proposed to use linear motors in pairs for a cardiac compression device element as shown in Fig. 1. The dark blocks are moving elements of the rectangular plate linear ultrasonic motors. Connecting them is a soft flexible belt. The housing supports the moving elements, and guides its movement. When the moving elements move toward each other, the belt loosens, and the cardiac compression device element relaxes. When the moving elements move away from each other, the belt becomes tight, and the cardiac compression device element contracts. It was noted that optimum heart muscle contraction is approximately 25% under normal condition [19]. This suggests that any potential motors should have a large displacement but in a limited housing space to achieve the required contraction for a cardiac compression assist device element. The rectangular plate linear ultrasonic motor has two kinds of the moving elements [17,18]. One is the self-moving, where the vibrator moves along a guide. The other one is the moving bar, where the bar is driven along a guide. In a limited housing, the displacement of the moving bar is, at most, one half of the length of the housing space. The maximum displacement for the self-moving vibrator, however, is the length difference between the housing and self-moving element. Since the length of self-moving elements can be smaller than the half-length of the housing space and the housing can be directly used as a guide without beam, a selfmoving rectangular plate linear ultrasonic motor housed in a

216 Y. Ming et al. / Sensors and Actuators A 119 (2005) 214 220 guide is chosen as the actuator for the cardiac compression assist device. 3. Design and analysis 3.1. Design of dimensions of piezoelectric ceramics To decrease the weight of the actuator, a piece of rectangular piezoelectric ceramic is directly used as the vibrator. Using finite element software ANSYS, the dimensions of the rectangular piezoelectric ceramics have been carefully chosen on the basis of modal analysis to make the first longitudinal resonant vibration and the second lateral bending resonant vibration occur at the same frequency. When the dimensions are 30 mm 8mm 2 mm, its longitudinal resonant frequency and the lateral bending resonant frequency are very close, 50 781 Hz and 50 549 Hz, respectively. Figs. 2 and 3 show the longitudinal vibration mode with resonant frequency of 50 781 Hz and the second lateral bending vibration mode with resonant frequency of 50 549 Hz. When working frequency is close to these two resonant frequencies, the superposition of the longitudinal and bending vibrations can produce elliptical motions on its four ends, which will drive the piezoelectric ceramics to move. 3.2. Motion analysis For an actuator for direct cardiac compression assist devices, the control of the motion direction is essential. Hence, the electrodes of piezoelectric ceramics have been carefully divided, as shown in Fig. 4, where (a) (d) are four equal electrodes on the upper surface of the piezoelectric ceramic, p1, p2, p3, and p4 are four tips of friction material for generating thrust force. Fig. 4. Patterns of electrodes on piezoelectric ceramics. Fig. 5. Vibration in x direction at p1 and p2. To analyse the motion of the motor, finite element software ANSYS is used to obtain the harmonic response when a sine wave power with voltage of 50 V and a frequency varying from 50 220 Hz to 51 000 Hz, is applied on electrodes (b and d) in Fig. 4. Because the rectangular plate vibrator is usually pressed against the guide or beam, the tips on one side of the vibrator are used to drive the motor. Therefore, calculated vibrations at p1 and p2 are shown in Figs. 5 and 6. Fig. 5 shows that the vibrations in x direction at p1 and p2 are in opposite phase when the working frequency is close to the longitudinal resonance, and the vibration amplitude in the x direction at p1 is greater than at p2. This is because the vibration at p1 contains the contribution of inverse piezoelectric effect whereas the vibration at p2 does not. Fig. 6 shows that vibration amplitude in y direction at p1 is greater than at p2 if the working frequency is larger than second bending resonant frequency, and the vibrations in y direction are Fig. 2. Longitudinal vibration mode at frequency of 50 781 Hz. Fig. 3. Lateral bending vibration mode at frequency of 50 549 Hz. Fig. 6. Vibration in y direction at p1 and p2.

Y. Ming et al. / Sensors and Actuators A 119 (2005) 214 220 217 When the piezoelectric ceramic extends in the x direction as shown in Fig. 7(a), vibrations in the y direction at p1 and p2 satisfy Y p1 > Y p2. Hence, from Eqs. (1) and (2), wehave F p1 >F p2 (3) This suggests that the tip p1 will stick to the guide surface while the tip p2 will move due to the extension of the piezoelectric ceramics in the x direction, shown as the dotted arrow at tip p2 in Fig. 7(a). When the piezoelectric ceramic contracts in x direction as shown in Fig. 7(b), the vibrations in the y direction at p1 and p2 satisfy Y p1 < Y p2. That is F p1 <F p2 (4) Fig. 7. The analysis of motion direction. in opposite phase when working very near to the bending resonance. Below the second bending resonant frequency, there exists a narrow frequency range where the vibration amplitude in y direction at p2 is greater than at p1. Since this narrow frequency range is just about 100 Hz, it will be difficult to change the motion direction through varying the frequency. However, it is found that the motion direction can be changed through applying voltage to different electrodes because the vibration amplitudes are greater at p1 than at p2 when the driving voltage is applied to electrodes (b and d) with the working frequency between the two resonant frequencies. Conversely when electrodes (a and c) are applied with driving voltage larger vibration amplitudes at p2 are induced, leading to an opposite motion direction. Details can be explained as below. As shown in Fig. 7, tips p1 and p2 are pressed against the surface of the guide. Assume that the pre-load is bigger than the y direction force produced by the friction material shape change at tips p1 and p2. Since the pre-load can be considered as equally applied on p1 and p2, F p /2 on each, the friction force between p1 and the guide surface can be written as: ( F p1 = Y p1 K + F ) p µ (1) 2 where Y p1 is the vibration in y direction at tip p1, K the elastic coefficient of the friction material, and µ is the friction coefficient. And the friction force at the tip p2 is: ( F p2 = Y p2 K + F ) p µ (2) 2 where Y p2 is the vibration in y direction at the tip p2. Assume that the working frequency of the motor is between the two resonant frequencies of the vibrator. Since the vibration amplitude in y direction at p1 is greater than at p2, we have Y p1 > Y p2 in the positive half cycle and Y p1 < Y p2 in the negative half cycle. Hence, the tip p2 sticks to the guide surface while the tip p1 will move due to the contraction of the piezoelectric ceramics in the x direction, shown as the dotted arrow at tip p1 in Fig. 7(b), that is, the motor moves in x direction. When an input voltage is applied on the electrodes (a and c), the tip p2 will have a larger vibration amplitude if the working frequency is same as before. The motor will move in the reverse direction based on the same analysis. Therefore, the direction of the motor is controlled through applying input power to different electrodes. Assume that the vibrator moves in x direction, the vibration in x direction at the tips decides the velocity of the motor to some extent, while the vibration in y direction at the tips mainly determines the thrust force. Since the second bending resonance is slightly different from the longitudinal resonance, the maximum thrust force and maximum output velocity of the motor will occur at different working frequencies. The maximum thrust force occurs near to the second lateral bending resonance because the vibrations in the y direction at p1 and p2 are in opposite phase, leading to a maximum difference between F p1 and F p2. Maximum output velocity occurs near the longitudinal resonance because of its relative larger vibration amplitudes in the x direction. 4. Experiment results A large size prototype of self-moving linear ultrasonic motor has been fabricated and housed in a guide as shown in Fig. 8. To increase the thrust force of the motor, two identical piezoelectric ceramics are fixed within aluminium structure, Fig. 8. Large size prototype motor in a guide.

218 Y. Ming et al. / Sensors and Actuators A 119 (2005) 214 220 Fig. 9. The schematic diagram for measuring resonant frequency. in which sponge materials are used to apply a pre-load. The whole linear ultrasonic motor weights 56 g with dimensions 43 mm 25 mm 3 mm. When input power is applied to the electrodes wired with solid line as shown in Fig. 9, the motor moves left and moves right in the guide when the electrodes wired with dotted line are connected to input power. 4.1. Resonant frequencies The ultrasonic motor is designed to work near resonance. Therefore, it is necessary to measure the resonant frequencies of the motor under actual operating conditions to obtain the best performance. A schematic diagram is illustrated in Fig. 9, where the oval shape E represents spring materials, in this case it is sponge. When a sine wave input is applied to the electrodes (b and d) of upper piezoelectric ceramic plate and (a and c) of lower one wired with solid lines, the remaining electrodes connected with dotted lines would sense the signal of the vibration. By adjusting the frequency of the generator, the output signal amplitude on an oscilloscope will change. Fig. 10 shows the ratio of output signal amplitude to its input amplitude with respect to frequency. In Fig. 10, there exists a minimum value at 51.18 khz, where a pure bending vibration occurs. The two pairs of electrodes (a and c) and (b and d) have opposite vibrations, resulting in a minimum output due to the counteracting piezoelectric effect. The actual second bending resonant frequency is 600 Hz above that suggested by the finite element calculation result. In Fig. 10, there is also a maximum value at 52.98 khz, which occurs due to the first longitudinal resonant vibration in which all parts Fig. 10. The ratio of output amplitude to its input vs. frequency. Fig. 11. Schematic diagram of experiment rig. of the vibrator expand and contract at the same time. This produces an intensifying piezoelectric effect, which increases the output to its maximum. The actual longitudinal resonant frequency is 2.2 khz higher than the calculated results. The holding force in the x direction causes this deviation of the longitudinal resonant frequency when the vibrators are fitted into the aluminium structure. 4.2. Frequency response characteristics From motion analysis based on finite element calculation, it is known that the motor will not work at the ideal condition. Its maximum thrust force and velocity will occur at different working frequencies. To characterise the performance of the motor, it is necessary to investigate its behaviour at different working frequencies. Fig. 11 shows the schematic diagram of the equipment used to test the performance of the linear ultrasonic motor. Here, a computer with an analogue to digital converter (ADC) board with LabVIEW software (National Instruments, Austin, USA) is used to control the system. The frequency signal can be generated from a signal generator or a counter in the ADC board. The signal generator produces a pure sine wave, which is used to test the performance of the motor. The counter in the ADC board can generate series pulses, which can be programmed by computer to control the motor. The direction control signal and the pulses first pass through control circuits, where the digital signals are insulated from the power amplifier and the voltage of the direction control signal is increased to drive the relay. The relay decides which electrodes are to be connected to the power amplifier. Hence, the motion direction is reversible. A spring is linked between linear ultrasonic motor and the force sensor, applying different forces to the linear ultrasonic motor when it moves. When the motor and spring is in balance, the maximum thrust force is obtained. A laser displacement sensor is used to measure the displacement of the motor without contact. The ADC samples both analogue signals from the force and laser sensor. A control and test program with fixed scan rate has been developed. The time interval between the sampled data is determined by scan rate. Therefore, the velocity of the linear motor can be acquired from the displacement variation divided by the time interval. Since the longitudinal resonance differs from the second lateral bending resonance, it is necessary to find the working

Y. Ming et al. / Sensors and Actuators A 119 (2005) 214 220 219 Fig. 12. Maximum thrust force vs. frequency. frequency where the maximum thrust force and velocity occur. With the input sine wave voltage amplitude fixed at 100 V, the working frequency is adjusted from 49.5 khz to 53.5 khz. The maximum thrust force and maximum velocity are recorded in Figs. 12 and 13. It is observed that the maximum thrust force occurs below 52 khz whereas maximum velocity occurs over 52 khz. The frequency difference is about 400 Hz. These results suggest that the maximum thrust force take place close to the second lateral bending resonance whereas the maximum velocity takes place near the first longitudinal resonance. A relative large maximum thrust force can be maintained from 51 khz to 52 khz, which reduces significantly when the working frequency is over 52 khz. Hence, the working frequency should be chosen according to the priority of the thrust force, the velocity, or a comprising between the two. For the direct cardiac compression assist device, it is hoped to obtain maximum thrust force and maximum velocity at the same working frequency. Therefore, reducing the frequency difference between the longitudinal resonance and bending resonance will be a key priority in the development of a self-moving rectangular plate linear ultrasonic motor for use as an actuator of the direct cardiac compression assist device. 4.3. Force control To be the actuator of the cardiac compression assist device, the motor should exert sufficient force to provide the Fig. 14. Thrust force vs. time. Solid line denotes desired force, dotted line denotes measured force. Fig. 15. Displacement vs. time under force control. level of assistance desired, however, the force needs to be carefully controlled so as not to damage the heart. Therefore, a force control of the motor is tested as a first step in evaluating the motor s performance. The thrust force of the motor is adjusted through modulation of the pulse width generated by the counter in the ADC board with a fixed working frequency of 51 khz. Using a PID algorithm, the pulse width is calculated from the subtraction between the desired force thrust force and the measured thrust force. When the motion period is 1 s and its duration for exerting force time is 30%, the motor s performance is showed in Figs. 14 and 15. Itis observed that thrust force can change from 0 to 68% of the desired force in about 100 ms after 8 mm displacement. The motor s performance is promising for being the actuator of the cardiac compression assist device. However, the maximum thrust force is smaller than the expected. The main reason is that the control strategy and the algorithm have not been optimised. The initial response delay between desired force and the measured force is mainly caused by the scan rate of the programme, which is limited by the hardware. Here scan rate is 50 s 1, resulting a 20 ms delay. 5. Conclusion Fig. 13. Maximum velocity force vs. frequency. To develop a real actuator for cardiac compression assist devices, there are still a lot of problems to solve. In this paper, a primary attempt to develop a self-moving linear ultrasonic

220 Y. Ming et al. / Sensors and Actuators A 119 (2005) 214 220 motor with large displacement in a limited housing space for use in cardiac compression assist devices has been made. The thin structure, lightweight, fast response, high force density, reversible motion, single power driving, and large displacement within a limited housing configuration of the rectangular plate linear motor are promising features to fit the needs of the cardiac compression assist devices. Using two pieces of piezoelectric ceramics, a self-moving rectangular plate linear ultrasonic motor has been fabricated and tested. Its second bending resonant frequency is about 600 Hz higher than the calculated result, its first longitudinal resonant frequency is about 2 khz higher than the estimated result. Its maximum thrust force reaches about 0.45 N, and maximum velocity 120 mm/s. Also under force control, 1 Hz pulse desired force is realised with thrust force changing from 0 to 68% of the desired force in about 100 ms after 8 mm of displacement. However, theoretical analysis and experimental results indicate that the maximum thrust force and maximum velocity do not take place at the same working frequency, resulting in a deteriorated performance. Therefore, reducing the frequency difference between longitudinal resonance and bending resonance is a key issue in improving the motor s performance and further investigation is necessary especially for the intended application. Acknowledgement The authors would like to acknowledge the support of the National Heart Research Fund, UK. References [1] G. Davies, M.C. Levesley, P.G. Walker, K. Watterson, Proceedings of the IEEE CCA/CACSD Conference, Glasgow, 2002, pp. 669 674. [2] D.J. Goldstein, M.C. Oz, Cardiac Assist Devices, Futura Publishing Company, New York, 2000, pp. 387 402. [3] E.A. Cheever, D.R. Thompson, B.L. Cmolik, W.P. Santamore, Stimulator for skeletal muscle cardiac assist, IEEE Trans. Biomed. Eng. 45 (1) (1998) 56 67. [4] J.H. Artrip, G. Yi, J. Shimizo, E. Feihn, R.R. Sciacca, J. Wang, D. Burkhoff, Maximizing menodynamic efffectiveness of biventricular assistance by direct cardiac compression studied in ex vivo and in vivo canine models of acute heart failure, J. Thorac. Cardiovasc. Surg. 120 (2000) 379 386. [5] M. Shahinpoor, Heart assist devices equipped with ionic polymeric platinum composite artificial muscle, ASME Bed. 36 (1997) 263 264. [6] T. Morita, Miniature piezoelectric motors, Sens. Actuators A 103 (2003) 291 300. [7] S. Ueha, Y. Tomikawa, Ultrasonic Motors Theory and Applications, Clartipon Press, Oxford, 1993. [8] H. Hirata, S. Ueha, Design of travelling wave type ultrasonic motor, IEEE Trans. Ultrason. Ferroelect. Freq. Control 42 (2) (1995) 225 231. [9] J. Satonobu, D.K. Lee, K. Nakamura, S. Ueha, Improvement of the longitudinal vibration system for the hybrid transducer ultrasonic motor, IEEE Trans. Ultrason. Ferroelect. Freq. Control 47 (2000) 216 221. [10] M.S. Tsai, C.H. Lee, S.H. Hwang, Dynamic modelling and analysis of a bimodal ultrasonic motor, IEEE Trans. Ultrason. Ferroelect. Freq. Control 50 (3) (2003) 245 256. [11] J.H. Green, Basic Clinical Physiology, Oxford University Press, 1981. [12] M. Kurosawa, S. Ueha, High speed ultrasonic linear motor with high transmission efficiency, Ultrasonics 27 (1989) 39 44. [13] T. Hemsel, J. Wallaschek, Survey of the present state of the art of piezoelectric linear motors, Ultrasonics 38 (2000) 37 40. [14] B. Zhai, S.P. Lim, K.H. Lee, S. Dong, P. Lu, A modified ultrasonic linear motor, Sens. Actuators 86 (2000) 154 158. [15] S. He, W. Chen, X. Tao, Z. Chen, Standing wave bi-directional linearly moving ultrasonic motor, IEEE Trans. Ultrason. Ferroelect. Freq. Control 45 (5) (1998) 1133 1139. [16] J.F. Manceau, F. Bastien, Linear motor using a quasi-travelling wave in a rectangular plate, Ultrasonics 34 (1996) 257 260. [17] Z. Jona, Ceramic motor, U.S. Patent 5777423 (July 7, 1998). [18] C. Li, C. Zhao, A large thrust linear ultrasonic motor using longitudinal and flexural modes of rod-shaped transducer, IEEE Ultrason. Symp. 1 (1998) 691 694. [19] G.L.J. Daves, Control of a Cardiac Compression Assist Device, Ph.D. thesis, September 2002.