ABSOLUTE CALIBRATION OF CONDENSER TRANSMITTERS 113 



means for keeping the relatively large driving voltage out of the 

 transmitter output circuit. 



In terms of Fig. 8 the sensitivity of the transmitter is given by 



i?=ivol,s/bar, . = ,..„.10-»«-. P = ,^,^^^,^.j,,. . 



where V^!2 = Vi, measured in volts, iaV2 = ii, h = separation 

 between AI and D. The e.m.f. e is measured by means of the potential 

 attenuator P, carrying a known current ia, and having an input 

 resistance /-„. At any one frequency two quantities must be measured: 

 V, say with an electrostatic voltmeter, and a, the setting of the 

 attenuator in decibels. The current ia must be known but can readily 

 be kept constant at all frequencies if a heterodyne oscillator be used as 

 the source. 



Two corrections must be applied. First, the auxiliary electrode is 

 perforated. Hence not all of its area is electrostatically effective. 

 Second, p in the above is the electrostatic force per unit area, rather 

 than the acoustic pressure on the diaphragm. The two are different, 

 in general, because of the acoustic load (Zd) on the front face of the 

 diaphragm. The value of Zd is affected by form of C, the auxiliary 

 electrode, and by the acoustic impedance beyond it in the chamber G. 

 It is best to have Zd as small and as free from reactance as possible. 

 This is accomplished by using stretched fine metal gauze. Copper 

 gauze, 300-inch mesh, is quite good. It terminates in the tube TT, 

 3>^-inch iron wall, which is filled with several layers of loose cotton 

 batting and hairfelt. The effectiveness of the arrangement was 

 judged by the fact that altering the size of G did not appreciably affect 

 the calibration. 



While the screen electrode provides a practically uniform electro- 

 static pressure over the surface of D, it is rather complicated to 

 compute the effective absolute values of h and of the electrostatic 

 area. This is more easily done by comparing it at low frequencies (say 

 at 100 c.p.s.) with a steel plate electrode in which the perforations 

 take up about 12 per cent of the total area. The surface facing D is 

 carefully machined so that h is uniform and known within less than 

 ± 2 per cent. This is for absolute values of h in the range 0.075-0.080 

 cm. The acoustic load which this electrode imposes on D, with G 

 removed, is negligible at low frequencies. A lower limit on the 

 electrostatic correction for the perforations is made by adapting the 

 calculation given by Maxwell (" El. Mag.," 3d ed.) for rectangular 

 grooves in one plate of a parallel plate condenser. The above value of 

 R is corrected to 



