SELECTIVE FADING OF MICROWAVES 77 



Fig. ()(0 illustrates the case of abnormal reflection from the water of 

 Raritan Bay on the Murray Hill path as indicated in Fig. 3(c). Here the 

 "normal" signal component is arriving at 0.1° above the line-of -sight 

 while another component, almost equal in amplitude, is arriving at the 

 \-ery bottom of the scan, about 0.8° below the line-of-sight. It is quite 

 probable that there have been times when this component was present 

 hut was outside the range of the scanning antenna. 



The mechanisms discussed in connection with Fig. 3(b) and 3(d) can- 

 not be demonstrated by photographs such as those just presented al- 

 though the angle-of -arrival radar was instrumental in furnishing the 

 clues to the phenomena. Due to the two-way attenuation of the radar- 

 corner reflector technique, the signal at these times rapidly falls below 

 the noise level of the receiver. For the same reason, it is not possible to 

 detect the extra signal components of small amplitude which were postu- 

 lated to account for the very deep fades sometimes observed under these 

 transmission conditions. 



FREQUENCY-SWEEP OBSERVATIONS 



Since most of the fading is due to interference between waves which 

 travel over different paths of, presumably, different lengths it was 

 realized that the fading was likely to be frequency selective. Just how 

 selective would depend on the relative lengths of the individual trans- 

 mission paths. The usual methods for determining path length differ- 

 ences are to use short pulses, or to sweep the freciuency. Since it was 

 likely that the path-length differences would be measured in feet rather 

 than yards, very short pulses or a wide frequency-sweep were required. 

 An oscillator^ was available whose frequency could be swept over the 

 licensed band of 500 mc between 3700 mc and 4200 mc. The frequency- 

 sweep experiment was set up on the Murray Hill-Crawford Hill path 

 for the summer of 1949. The following summer, the milli-microsecond 

 pulse transmission tests described in the companion paper were con- 

 ducted over the same path. As might be expected, simultaneous observa- 

 tions showed good agreement between the two methods. 



The frequency of the transmitter, located at Murray Hill, is swept 

 over a 450-mc band centered at 3950 mc at a 60-cycle rate. At the 

 receiver, a similar oscillator is used for the beating oscillator except that 

 its frequency is swept linearly through the same frequency band in one 



^ This oscillator was developed b}- M. E. Hines and is described in his paper 

 published in the Bell System Technical Journal, Vol. 29, Oct. 19.50. It uses a 416A 

 close-spaced triode in a wave-guide cavity. The frequency is changed by means 

 of a plunger which is capacity-coupled to the plate of the tube and which is actu- 

 ated by a modified loud speaker unit. 



