38 



density of water. Since the wires used in this qualitative survey on the 

 tanker model were not cleaned, it is reasonable to conclude that their sensi- 

 tivities diminished during the experiment. In view of their relatively high 

 initial sensitivities, however, such attenuations as might have occurred were 

 not apparent in the response. A given wire reproduced qualitatively the same 

 oscillogram repeatedly under the same test conditions. 



FREQUENCY RESPONSE 



• The frequency response of the hot-wire circuit illustrated in Figure 

 2a depends not only on the response characteristics of the hot wire itself, 

 but also on the characteristics of the amplifier and recording galvanometer. 

 The frequency response of a hot wire is dependent upon its thermal 

 inertia or so-called thermal lag. This wire time constant is given by 



M = kms(R " R a } [28] 



i 2 R R a 

 a o 



where k is the mechanical equivalent of heat, 

 m is the mass of the wire, and 

 s is the specific heat of the wire. 



The other symbols are as defined for Equation [26] on page 36. If the ampli- 

 fier employed with the wire cannot compensate for this lag, the response to a 

 harmonic velocity variation of constant amplitude will be such that the ratio 

 of the amplitude of the velocity change at frequency f to the amplitude at 

 frequency zero is given by 



dE 1 



dE )^TVf¥ 



[29] 



The value of M, in water, has been measured and found to lie in the range from 

 70 to 100// sec. Consequently, a harmonic velocity variation of a given ampli- 

 tude would be reduced to half value at about 4000 cps compared to Its ampli- 

 tude at zero frequency. 



The frequency response of the TMB Type-3A amplifier used in this ex- 

 periment is constant from about 2 cps to well above 10,000 cps. On the other 

 hand, the galvanometers of the 14-channel Consolidated oscillograph attenuated 

 signals greater than 400 cps. In view of this instrument limitation compensa- 

 tion of the amplifier for thermal lag similar to that described in Reference 

 2 was unnecessary. 



