WIDE BAND TRANSDUCERS FOR SOUND VELOCIMETERS 



C. L. BUCHANAN 



U. S. Naval Research Laboratory 



Washington, D. C. 



ABSTRACT 



Some influences and considerations which play 

 a significant role in the accurate measurement of 

 underwater speed of sound are presented. An inac- 

 curacy of "sing around" circuits used in the mea- 

 surement of the speed of sound due to variations 

 in the measurement of the acoustic pulse travel 

 time is discussed. The contribution to the devel- 

 opment of an accurate velocimeter by the use of 

 proper amplifier bandwidth, pulse rise time, 

 transducer "Q" and threshold triggering level is 

 emphasized. 



INTRODUCTION 



The accurate measurement of sound speed is 

 the keystone of efforts to describe the various 

 acoustic paths in the ocean by ray plotting tech- 

 niques. The importance of sound speed measure- 

 ment is well recognized in qualitative terms but 

 the degree of accuracy required and the quanti- 

 tative effect of measurement error is not fully 

 appreciated. There is a real and immediate need 

 for sound speed measurements with a repeatability 

 of 1 part in 10,000 over the entire spectrum of 

 ocean environment . 



Several years ago the National Bureau of 

 Standards developed the "sing around" type veloc- 

 imeter. This development was an excellent 

 example of the kind of imaginative instrumenta- 

 tion which is required if we are to make real 

 improvement in our oceanographic measurement capa- 

 bility. On the other hand, there are basic con- 

 siderations with regard to measurement accuracy 

 that need to be reviewed with the objective of 

 improving this instrument to its ultimate capa- 

 bility. The paper discusses the capability of 

 current instruments, the factors which limit 

 their accuracy and the areas in which improve- 

 ment is necessary. 



BASIC DESIGN CONSIDERATIONS 



Pulse Travel and Rise Time 



The basic principle of operation of pulsed 

 velocimeters is measurement of the time required 

 for an acoustic signal to travel a fixed distance. 

 In the "sing around" system each received pulse 

 triggers a succeeding pulse so that the time 



TIME 



Lo tm tm' 



t = TIME OF INITIATION 

 t, = TIME OF COMPLETION 

 t m =TIME MEASURED 

 t m ,= TIME MEASURED 



WHEN GAIN CHANGED 

 At TIME ERROR 



DUE TO GAIN CHANGE 



Fig. 1. Pulse time measurement. 



measurement is accomplished by counting the number 

 of pulses occurring in a given time interval. 



Now just how does one measure the time of 

 occurrence of a transient event? Fig. 1 shows a 

 typical level shift function which is initiated 

 at time t and stops at time t~. The determina- 

 tion of the time of the pulse is made by observing 

 the time at which the transient voltage exceeds a 

 threshold voltage, E+ . 



The design of a velocimeter for any specified 

 accuracy puts a minimum requirement on the per- 

 missible error in the fundamental factors involved 

 in the time measurement. These are the rise time, 

 the threshold level and the loop gain of the sys- 

 tem. In Fig. 1 it should be noted that any shift 

 in the gain of the system will result in a shift 

 in the time indicated. The dashed curve repre- 

 sents the function reduced by a loss of 3 db and 

 the point, t' , represents the new time indicated. 

 The difference between the time, t m , and time, 

 t m , is the resulting error, t. By similar reason- 

 ing it can be shown that any shift in the thresh- 

 old level likewise produces a change in the indi- 

 cated time and therefore an error in time measure- 

 ment . One way to escape from this limitation is 

 the use of a level shift that is so rapid that 

 measurement at any place on the wave front yields 

 the desired accuracy. 



A design based on this philosophy could be 

 based on a path length of 7-6 inches and a maxi- 

 mum design error of ±0.5 feet per second (fps). 

 Here it is assumed that the path length is the 

 same as in the original "sing around" developed 

 by the National Bureau of Standards, and the 

 accuracy desired is equivalent to roughly 1 part 



157 



