STEADY STATE LOUD SPEAKER MEASUREMENTS 145 



ments on such devices at very low frequencies are consequently more 

 indicative of the capabiHties of the loud speaker when obtained 

 outdoors or in a very large room. 



Experimental Data 



General. — To determine the extent to which the above discussed 

 acoustic effects may influence the result of a loud speaker performance 

 measurement and to show the measuring conditions under which such 

 acoustic factors are encountered, response-frequency characteristics of 

 loud speakers were measured under various acoustic conditions. 

 These are described in the following paragraphs. 



In all these measurements a calibrated condenser transmitter 

 approximately 2|" in diameter was used as the acoustic detector. The 

 thermophone calibration on this transmitter showed that with constant 

 pressure on the diaphragm the voltage produced between the grid and 

 filament of the associated vacuum tube was very nearly independent of 

 frequency. Corrections for such small variations as did exist however 

 have been made in all the following curves. In addition a tapered 

 correction of .6 TU per 100 cycles increased in frequency between 

 1,100 and 2,100 cycles and a constant correction of 6 TU for fre- 

 quencies above 2,100 cycles have been subtracted from the response 

 measurements to correct for the fact that the indicated pressure 

 approaches twice the free space pressure at frequencies at which there 

 is reflection from the diaphragm. This latter correction was obtained 

 by making response-frequency measurements on a loud speaker under a 

 fixed set of conditions; first, with the condenser transmitter freely 

 suspended in the sound field as in all the following curves and then 

 with the transmitter located at the center of a round baffle 12" in 

 diameter. When in the baffle complete sound reflection from the 

 transmitter occurred at a frequency lower than that at which reflection 

 began to take place from the transmitter alone. From the difference 

 between the two response-frequency curves so obtained, it was there- 

 fore possible to definitely locate the transition frequency range between 

 1,100 and 2,100 cycles and to evaluate the transmitter reflection 

 correction. 



The number of measurements made in order to define any response- 

 frequency characteristic depended upon the nature of the curve. If 

 no abrupt changes in the response were observed in making the 

 measurements, approximately 10 measurements per octave were 

 obtained. Otherwise the frequency of the oscillator was changed by 

 small steps and a sufficient number of measurements made to clearly 

 define the curve. 



