A HIGHLY ACCURATE ULTRASONIC MEASUREMENT SYSTEM FOR TREMOR USING BINARY AMPLITUDE- SHIFT-KEYING AND PHASE-SHIFT METHOD

Transcription

1 APPLICATIONS, BASIS & COMMUNICATIONS 61 A HIGHLY ACCURATE ULTRASONIC MEASUREMENT SYSTEM FOR TREMOR USING BINARY AMPLITUDE- SHIFT-KEYING AND PHASE-SHIFT METHOD MENG-HSIANG YANG 1, K. N. HUANG, C. F. HUANG, S. S. HUANG 1 AND M. S. YOUNG 1 1 Department o Electrical Engineering, National Cheng-Kung University, Tainan, Taiwan Department o Electronic Engineering, I-Shou University, Kaohsiung, Taiwan Department o Electrical Engineering, Chengshiu institute o Technology, Kaohsiung, Taiwan ABSTRACT A highly accurate Binary Amplitude-Shit-Keyed (BASK) ultrasonic tremor measurement system or use in isothermal air is developed. In this paper, we present a simple but eicient algorithm based upon phase shits generated by three ultrasonic waves o dierent requencies. By the proposed method, we can conduct larger range measurement than the phase-shit method and also get higher accuracy compared with the time-o-light (TOF) method. Our microcomputer-based system includes two important parts. One o which is BASK modulation signal generator. The other is a phase meter designed to record and compute the phase shits o the three dierent requencies and the result motion is then sent to either an LCD or display or a PC or calibration. Experiments are done in the laboratory using BASK modulation or the requencies o Hz and 1 khz with a 4 khz carrier. The measurement accuracy o this measurement system in the reported experiments is within +/-.98 mm. The main advantages o this ultrasonic tremor measurement system are high resolution, narrow bandwidth requirement, low cost, and easy to be implemented. Biomed Eng Appl Basis Comm, (April); 15: INTRODUCTION Tremor, an involuntary movement, is characterized by regular or irregular oscillations o one or several segments o human body. It is resulted rom alternating or simultaneous contractions o agonist and antagonist muscle groups [1,]. The inherent oscillatory behavior o tremor allows characterization in terms o requency and amplitude. The requency is associated with the nature o the underlying mechanisms, while the amplitude determines symptomaticity. Tremor occurred on human bodies can be measured by many methods just like vibration measurement Received: March 1, ; Accepted: April 1, Correspondence: M.S. Young, Ph.D., Proessor Department o Electrical Engineering, National Cheng-Kung University, Tainan, Taiwan msyoung@mail.ncku.edu.tw in industrious ields. For instance, the methods such as strain gauges and accelerometers are developed to measure the vibration o an object. However, it is noted that strain gauges are quite sensitive and require high gain ampliiers, thus making susceptible noise problem []. Moreover, it is reported that the attitude o an accelerometer in a gravitational ield can cause incorrect amplitude measurement [4]. Furthermore, they are not convenient in some ways where the measuring devices can not be attached directly onto the vibrating structure. Non-contact methods or remote measurements o vibration are also developed. The most sensitive ones are based on optical intererometry. However, optical techniques are generally sensitive to smooth surace and the devices based on these techniques are expensive [5]. The other method is based on the ultrasonic techniques. Most o them measure the time-o-light (TOF) o a pulse as it travels between the transmitter and the receiver, and then multiply the time by the sound speed to get the distance between transducers -15-

2 6 and object [6-9]. However, the measurable vibration requency range o an object decreases when the distance between the object and transducers increases. To overcome the drawback, Figuroa and Barbieri oered an ultrasonic ranging system based on the phase-shit method or structural vibration measurement [7-1]. However, the measurement resolution varies with the range o the displacements being measured. In other words, the larger the measurement range is, the lower the resolution readout will be. In this study, a new method is developed to improve the ormer disadvantages. This system is based upon the transmission o a Binary Amplitude-Shit- Keyed (BASK) modulated signal. There are two steps or this system to measure the motion o an object. First, the phase shit is determined by the envelopes o the transmitted and the received signals. Second, the carrier phase shit between transmission and reception is also computed in order to reine the inal result. In this system, a BASK modulation is applied by using low-cost ultrasonic transmitters and receivers, which have high selectivity and hence narrow bandwidth. The phase shits are computed by using a counter technique to avoid the limitation caused by the amplitude o the signal and the inite bits o the A/D converter. The waveorm representing the relative motion o an object can be measured rom the phase dierence variation between the continuously transmitted and relected ultrasound.. METHOD Vol. 15 No. April As shown in Fig. 1, two ultrasonic transducers, one transmitter and one receiver, are attached together in the non-contact ultrasonic tremor measurement system. For the moving object, the angle between the incident ultrasonic wave P i ( and the relected wave P r ( is negligible. In operation, the ultrasonic transmitter transmits P i ( continuously and the ultrasonic receiver will simultaneously receives P r ( relected by the object. The transmitted and the received waveorms are given by P i (=A i sin(ω, (1) P r (=A r sin(ωt-l(ω/c), () where ω is the angular requency o the waveorm, c is the speed o sound, and L( is the distance between the ultrasonic transducers and the object. The actor in the P r ( unction represents that the travel path o the ultrasound is two times o L(. Thus, the phase delay between the transmitted and the received waveorms is given by Φ( = (/λ)*l(, and the ormula can be written as θ ( λ d ( = L( =, () where λ is the wavelength, θ(= Φ is the phase shit and its value usually cycles between zero to. The moving distance, d(, can be uniquely determined by the phase shit θ( and the maximum ranging distance does not exceed the hal wavelength. I the maximum ranging distance exceeded the hal wavelength, a phase ambiguity occurs. In other words, the measured moving distance is limited to hal wavelength (λ/). For instance, λ is approximately 8.96 mm when a 4 khz signal is used. When the movement o an object is more than 4.48 mm (λ/), the value o θ varies around. Such a small maximum range greatly limits the use o phase detection methods to measure moving distance. To overcome this limitation, a method to count the integer number o wavelength occurring as the measured displacement increases is used. The total displacement D( is determined by the ormula, θ ( D ( = d( = n λ + λ, (4) where n is the number o the wavelength. Even though many methods have been discussed in recent literatures, we develop a novel method to overcome the limitation. The ranging principle is similar to the operation o a ruler. Three dierent acoustic signal requencies, 1, and are used to measurement the total displacement D(, where 1 < < as L Ai t Transmitter d Ar Φ(=L(ω/c t Receiver Fig. 1. The non-contact tremor measurement by adopting two ultrasonic transducers -16-

3 APPLICATIONS, BASIS & COMMUNICATIONS 6 θ1 θ θ D D1 λ1 N λ N λ Fig.. The motion distances are measured with three dierent requencies 1 1 (b) Fig.. (a) Three dierent requency signals (b) Signals modulated by BASK method shown in Fig.. For instance, the phase shit θ 1 o the 1 signal with the longest wavelength λ 1, gives the irst approximation o the displacement D 1 i the movement θ1( c θ( c θ( c = Int [ ] + Int[ ] + o the object does not exceed hal wavelength. 1 θ1( D ( D1 ( = λ1. (5) Using a higher requency signal to measure the same range with a iner ranging resolution causes a greater phase shit than requency 1 signal due to the small change in the pathlength. The integer number N o wavelength o requency signal can be calculated rom N = Int[D 1 / λ ]. The second approximation to the displacement D( is obtained rom the phase shit θ o requency signal, plus an integer number N o the wavelength λ, as described in the ollowing equation: θ( D ( N λ + D = N λ + λ,where θ ( D = λ, Similarly, the phase shit θ o requency signal, and N = Int[D / λ ] D ( N λ + N λ + D, where θ( D = λ, (6) The inal equation concluded rom the above equations is shown as ollows, θ( D ( = Int[ D1 / λ ] λ + Int[ D / λ] λ + λ θ1( λ1 θ ( λ θ ( = Int [ ] λ + Int[ ] λ + λ λ λ (a) (7) where c is the speed o sound. The maximum distance and minimum resolution whether can be achieved or not are determined by the choice o requencies 1, and. In order to be able to measure large displacement, requency o Hz is selected because it has the wavelength o approximately 17 mm in air. Frequency 4 khz is also selected because most o the available ultrasonic transducers are o low-cost. Furthermore, an intermediate requency o 1 khz is chosen in our experiment. Most available ultrasonic transducers cannot transmit the three ormer selected requencies without severe attenuation o one or more o these requencies because o their high selectivity. One solution to solve this problem is to use amplitude modulation [11]. Several low requency signals are applied to modulate the resonant requency signal o the transducer, and the phase shit o the envelope waveorm o the received signal is measured. In that study, three requencies (4 khz, 4 khz, 4 Hz) are adopted and a maximum detectable range about 41 mm is achieved, whereas the bandwidth o the transducers is about 8 khz. The system is easily implemented even by using some inexpensive transducers., A novel method is developed by us and it is more suitable or the microcontroller-based system with 4 khz o pulse signal to energize transducers. The low requency pulse signals are used to modulate the 4 khz pulse signal by BASK modulation. For example, only the high level part in the 1 khz pulse signal is used to modulate ten 4 khz pulse signals, as de- -17-

4 64 Vol. 15 No. April Signals Source Transmitter d Digital Phase Meter Bandpass Fillter and Comparator Preampliier and AGC Ampliier Receiver LCD 89C51 Microcomtroller Calibration System Fig. 4. Block diagram o our ultrasonic tremor measurement system 4 MHz OSC Divide 1 Counter 4 khz signal Divide 4 Counter 1 khz signal Divide Counter Hz signal AND Gate Fig. 5. The crystal oscillator circuit which is used to generate stable signals and then all the signals are modulated by an AND Gate scribed in Fig.. When the transmitted signal being received, the envelope o the received signal is picked up to reconstruct the pulse signal without the carrier. By applying this method, the three dierent pulse signals are gotten at the same time. From the detection o the phase shits by comparing the reconstructed signal with its corresponding original 4 khz, 1 khz or Hz signal, the moving distance is thus calculated.. SYSTEM IMPLEMENTATION As shown in the block diagram o Fig. 4, our ultrasonic tremor measurement system consists o two ultrasonic transducers (one o them is used to transmit the signal and the other is used to receive the relected signal), a signal source system, an auto-gain-controlled ampliier system, bandpass ilter and comparator, digital phase meters, and a calibration system. A microcontroller is used to govern the operation o the entire system..1 Hardware o the System An ultrasonic tremor measurement system can be conveniently divided into six parts. The operation o each module is described as the ollows Ultrasonic Signal Source A programmable logic device FPGA FLEX1 K1 (Altera, USA) is adopted in our system. The FPGA s gate coniguration is designed by the Verilog hardware description language (Cadence, USA). As shown in Fig. 5, a quartz crystal oscillator circuit is used to generate a stable 4 MHz signal. One divider is set at 1 to divide the 4 MHz signal to produce a 4 khz stable signal or our ultrasonic transmitter. By using the same method, a 1kHz and a Hz stable signals can be produced. Then, the three signals are modulated by an AND gate. Finally, the TTL level modulated signal is connected to the operational ampliier to generate a ± 1 V output signal or our 4 khz ultrasonic transmitter..1. Auto-Gain-Controlled (AGC) Asmpliier Acoustic attenuation eects are large when the distance between transmitter and receiver is over 5 cm. -18-

5 APPLICATIONS, BASIS & COMMUNICATIONS 65 Temperature Controlled Room Motor Controller Stepping Motor Brass Plate Ultrasonic Transmitter represents the phase dierence. Three counters based on 4 MHz oscillator are adopted to count the pulse width at dierent requencies. One divide-by-1 counter, the other divide-by-4 counter and another divide-by- counter are used to detect the phase dierences with a resolution o /1, /4 and / radians. Finally, the counting values are sent to the 89C51 microcontroller. RS PC RS Optical Linear Scale Ultrasonic Tremor Measurement System Fig. 6. The experimental structure o our tremor measurement system To reduce the errors rom acoustic attenuation, the gain o the ampliier must be adjusted automatically when acoustic attenuation occurs. Thereore, the error eects related to acoustic attenuation will be minimized in this AGC module by keeping the amplitude o the received signal dynamically constant. However, the phase o the received signal is dependent on the gain o the ampliier. During the calibration, phase error such as a unction o AGC gain is measured and recorded or the AGC ull gain range at a known distance..1. Bandpass Filter and Comparator Since the received signal is a modulated signal, an envelope detection circuit is required to demodulate the received signals. A bandpass ilter is used to ilter the high requency harmonics and the DC term rom the received signals. Separate sets o the bandpass ilters to center requencies o Hz and 1 khz are also used. The iltered signals are then converted by a voltage comparator into square waves. The square waves are converted into TTL-compatible signal by a -5 V limiter circuit, which is connected to digital phase meters..1.4 Digital Phase Meter The same FLEX1K1 chip is adopted or implementing the digital phase meters. The transmitted signals o 4 khz, 1 khz and Hz requencies are also used as the reerence signals when the phase shits are measured. The transmitted and the received signals are compared in a D-type lip-lops whose output pulse width LCD.1.5 Microcontroller The system is governed by an 89C51 (Atmel, USA) microcontroller. Ater receiving three phase shit values rom the FPGA chip, the movement distance o an object is calculated by Eq. 7, and then displayed on a LCD module. For advanced analysis, the experimental data can be transmitted to a PC through the RS- serial port and all the data are then recorded and displayed on the PC..1.6 Calibration System The ultrasonic tremor measurement system must be calibrated with a known distance. The calibration procedure has to be done at a constant temperature over the entire range. As shown in Fig. 6, the experimental embodiment o the ultrasonic tremor measurement system consists o a constant temperature chamber, a brass plate relector, a stepping motor with an optical linear scale, and a PC. The reerence scale o an optical linear measuring device (Pulscale model SJH5515AAR, Futaba, Japan) is ixed on the stepping motor. The stepping motor can be controlled to change the position o the relector mounted on the arm o the stepping motor. In this system, all the positions are set by PC and then the phase shit values and optically measured distance are provided. Thus, the error o the movement is calculated and displayed on PC. A lookup table is then made and all the errors o the movement are perormed. The calibration sotware used in PC is written by LabVIEW 6. (National Instruments, USA). The main unction o the sotware is to get values o phase shit rom the 89C51 microcontroller. The values o the phase shit and moving distance d( is then displayed on the screen.. Sotware o the System The 89C51 microcontroller is programmed to assign the transmitted signal, adjust the gain-controlled ampliier, and wait to be interrupted by the digital phase meter to evaluate the phase shit, get the phase shit θ 1, θ, θ and calculate the motion o the object. During the phase calibration, the phase shit, θ 1, θ and θ, are sent to the PC via the 89C51 microcontroller s RS interace. Ater calibration, the values o the measured distance and vibration are displayed on the LCD module. -19-

6 66 Vol. 15 No. April ERROR Error Distance (mm) 數列 1 Actual Position (mm) Fig. 8. The error plot o distance Fig. 7. The operation graph o the object at the position at 5 mm with 1 mm vibration amplitude 4. TESTING THE SYSTEM 4.1 Experimental Method A prototype o our ultrasonic tremor measurement system is tested in the laboratory by comparing the calculated distance obtained rom our system and the motion measured by the optical linear scale. An experimental system is established just as that described above. The center requency o the transmitted/received transducer is 4 khz. A one-dimensional ball screw with the relector mounted above connects with a stepping motor. The stepping motor has 4 steps per circle and the resolution o the ball screw is 1 µm per step, i.e. the lateral motion is 4 mm per screw revolution. Net transducer lateral travel speed during calibration is programmable rom 14 to 8 µm/s. The whole system is operated in a large plastic box. The inside wall o the box is covered by sponge which is used to absorb the ultrasound energy and avoid the airlow intererence. All experiments are preceded in a laboratory at 7 C controlled by the air conditioner. An oscilloscope (Agilent 546D, USA) is connected in parallel with the transducers to monitor the phase shits. 4. Results Figure 7 shows the operation graph o the object at the position at 5 mm with 1 mm vibration amplitude which are programmed by the sotware written by LabVIEW. The waveorms show the moving distance o the object ater the calculations o the three obtained phase shits in PC. The moving distance o the object is controlled rom 5 mm to 75 mm with dierent vibration amplitudes by the increase o the distance at 1 mm every time. The experiments are handled 1 times at every moving distance. The error data in the error plots is shown in Fig. 8. The standard error value is the statistic result o the repeating test or 1 times in each condition. The standard errors o the measured results are calculated by the ollowing equation, Standard Error = n i= 1 [ RP( i) OP] / n where RP is calculated rom the ultrasonic tremor measurement system, OP is the position o the moving object measured by an optical linear scale, and n is the number o measured data points. The average o the standard error is about.98 mm rom the obtained data in our experiments. 5. DISCUSSION A new high accurate ultrasonic tremor measurement system or use in isothermal air is presented. The system utilizes the BASK modulation methods to modulate dierent requency signals and obtain dierent phase shits to reconstruct the relative motion o the object. The phase shits are computed by a counter technique to avoid the limitations caused by the amplitude o the signal and the inite bits o the A/D converter. Our system being tested proves that it can monitor tremor occurred on human hands in the laboratory. By applying this ultrasonic tremor measurement system, we must consider the temperature and the pressure which aect the measured result. These variations change the speed o the sound in the air. Thereore, including a temperature compensation method to calculate the accurate speed o sound can improve the perormance o the ultrasonic measurement system. REFERENCES 1. C. D. Marsden, Origins o normal and pathological tremor, in Movement disorders: Tremor, J. Findley and R. Capildeo, Eds. London, England: McMillan, 1984; R. R. Young, Tremor, in Diseases o Nervous --

7 APPLICATIONS, BASIS & COMMUNICATIONS 67 System. Clinical Neurobiology, A. K. Asbury, G. M. Mckhann., and W. I. McDonald, Eds. Philadelphia, PA: Saunders, 1986; 1: J-N. Juang, L. G. Horta and H. H. Robershaw, A slewing control experiment or lexible structures, J. Guidance and Control, 1986; 9: 5: M. A. Norris, R. C. Thompson, and A. Das, Low- Frequency response o accelerometers or observer design in gravity environment, J. Guidance, 199; 1: 4: J. P. Monchalin, :Optical detection o ultrasound, IEEE Trans. Ultrason., Ferroelect., Freq. Contr., 1986; : C. Claudio, C. D. Giorgio, M. Brouno, P. Maria, and T. Andrea: A temperature compensated ultrasonic sensor operating in air or distance and proximity measurements. IEEE Trans. Instrum. Meas. 198; 9: F. Figueroa and E. Barbieri: Increased measurement range via requency-division in ultrasonic phase detection methods. Acustica 1991; 7: F. Figueroa and E. Barbieri: An Ultrasonic Ranging System or Structural Vibration Measurements. IEEE Trans. on Instrumentation and Measurement 1991; 4: 4: M. S. Young and Y. C. Li: A high precision ultrasonic system or vibration measurements. Rev. Sci. Instrum. 199; 6: 11: O. B. Matar, et al.: Noncontact measurement o vibration using airborne ultrasound. IEEE trans. On ultrasonics, erroelectrics, and requency control 1998; 45: : A. K. Lee, J. Lucas and L. E. Virr: Microcomputercontrolled acoustic rangeinding technique. J. Phys. E: Sci. Instrum. 1989; :