«i 



TESTING TECHNIQUE 



basis, which may be accomplished by making the 

 source of electrical power have a high output im- 

 pedance compared to the impedance of the crystal. 



Wide-Band Uniform Response 



It is desirable that a standard have a smooth fre- 

 quency response over a wide frequency band. Rapid 

 variations in response with frequency make it diffi- 

 cult to compare the responses of two instruments in 

 such a range because of the difficulty of reading ac- 

 curately a steep curve on a recorder chart. A wide 

 frequency band is desirable so that the number of 

 standards required to cover the entire frequency 

 range of interest be as small as possible to reduce 

 rigging time. For high-frequency transmitters it is 

 difficult to obtain a flat frequency response. However, 

 a smoothly varying response is in general satisfactory 

 and can be obtained readily with crystal projectors. 



Linearity, Large Dynamic Range 



Over the range of pressures which a standard is re- 

 quired to measure or to produce, the device should 

 be linear; that is, the voltage produced by a receiver 

 at each frequency should be proportional to the pres- 

 sure of the sound field in which it is contained, and 

 the pressure produced by a transmitter at some point 

 in the field at each frequency should be proportional 

 to the input current or voltage applied to it. When 

 nonlinearity occurs, a sinusoidal input signal (elec- 

 tric or acoustic) no longer, in general, produces a 

 sinusoidal output of the same frequency. Most instru- 

 ments, however, are linear over a limited range of 

 input signal amplitudes. For the greatest usefulness 

 of a standard, this dynamic range should be as great 

 as possible, as this allows a large range of sound field 

 pressures to be produced or measured by a single in- 

 strument. Since a transmitter may have variations in 

 response with frequency of 50 db or more, in order 

 to make a comparison calibration with a single stand- 

 ard receiver, the standard must have a corresponding 

 dynamic range as limited by nonlinearity on the high 

 end and inherent noise on the low end. 



Low Threshold 



The lowest pressure that can be measured by a re- 

 ceiver is determined by its inherent noise voltage. 

 This may be due to thermal noise, vacuum-tube 

 noise (of an associated preamplifier), contact noise, 

 or other similar factors. At each frequency, the pres- 

 sure at which the signal voltage of the instrument is 



ecpial to the noise voltage, in a 1-c band centered at 

 the frequency, is known as the threshold of the trans- 

 ducer. It is desirable to have the threshold of a stand- 

 ard as low as possible in order to extend the dynamic 

 range as far as possible in the direction of low pres- 

 sure. 



Reasonably High Response 



The magnitude of the response of a standard can 

 be of considerable importance independent of its 

 inherent noise characteristic. For example, if the re- 

 sponse is low, the electric crosstalk between a receiver 

 and a transmitter may exceed the level of the elec- 

 trical signal to be measured. A similar situation ap- 

 plies with respect to the use of a transmitter as a 

 sound source. Because of the presence of ambient 

 noise in the water, it is necessary that the response of 

 a transmitter standard be sufficiently great so that its 

 sound field exceeds the ambient noise sound field. 



A variety of standard transducers have been devel- 

 oped which satisfy the conditions outlined above. 

 Their characteristics are given in another part of 

 this volume. 



5-6 CALIBRATION OF DEVICES 



COVERING WIDE FREQUENCY RANGES 



When a transducer is being used lor wide-band 

 reception, a single frequency calibration is still sig- 

 nificant, because, if the device is linear, its wide-band 

 response can be determined from the single fre- 

 quency response by superposition in accordance with 

 Fourier's theory. The calibration then consists in de- 

 termining over what range the hydrophone is linear 

 and in taking a single frequency characteristic within 

 that range. The procedure for taking a single fre- 

 quency characteristic has been described. The linear- 

 ity at any given frequency is best observed by varying 

 the input level at that frequency and seeing whether 

 or not the output level follows proportionately. In- 

 stead of taking a single frequency characteristic, it 

 is, of course, also possible to measure the response for 

 a signal with any type of frequency spectrum. 



In particular, the response may be measured for a 

 signal consisting of very complex aperiodic wave 

 forms. Such signals usually are referred to as noises. 

 The sound created by thermal agitation, the so-called 

 thermal noise, is an illustration in point. 



The measurement of such signals places quite 

 severe requirements on the measuring system. It is 



