SELECTIVE FADING OF SUPER-HIGH FREQUENCY SIGNALS 1191 



path loss.^ In each record one second was required for the trace to travel 

 through the entire frequency range covered. 



The three records shown are consecutive, being taken at 3:05:21 AM, 

 3:05:25 AM and 3:05:27 AM. The first record (frame number 9524) 

 was taken with 10 db more attenuation in the input to the measuring 

 equipment than when the later two records were taken (frame numbers 

 9525 and 9526). There is obvious overloading in the left hand portions 

 of the records on frames 9525 and 9526; but the accuracy of the right 

 hand portions of these records (showing the deeper part of the fade) is 

 unimpaired. The noticeable difference between the shapes of the curves 

 near the deeper parts of the fade on frames 9525 and 9526 is typical. 

 These changes occurred within two seconds. Generally, the deeper parts 

 of the fades show more rapid changes than the less deep parts. 



From the 50,000 record photographs all those pertaining to fading 

 of 30 db or more were segregated and analyzed. The remaining records 

 were analyzed on a sampling basis, except that every record showing 

 unique effects was analyzed. About 1,800 path-loss versus frequency 

 curves such as those illustrated in Figs. 2 and 3 were obtained. These 

 curves were separately studied, and were also treated statistically. 



Because the levels during the deeper part of the fade are too low to 

 show on frame 9524, and because portions of frame 9525 showing the less 

 severe part of the fade are affected by overloading, it was necessary to 

 combine two records (shown on Fig. 1) to obtain the single path-loss 

 versus frequency curve shown as Fig. 2(a). 



Theoretically, it should be possible to synthesize by means of an addi- 

 tion-of- vectors method each of the path-loss versus frequency curves 

 obtained from these tests. Each vector term of the equation would 

 correspond to a component of the received signal and would be of the 

 form R cos coT, where: R is the magnitude of a particular component 

 (normalized to the magnitude of the direct signal component), T is the 

 delay (in seconds) between the time of arrival of the direct signal com- 

 ponent and the particular interfering component, and oj is 27r times 

 the frequency. 



In practice, however, it has been found in the case of deep fades that 

 components of relatively small magnitude and relatively long delay are 

 of importance in determining the shapes of the curves near maxima of 

 path loss. Also it has been found that in most such cases quite a few 

 components are involved. These factors make accurate analysis of many 



2 A logarithmic amplitude characteristic (db scale) would have been preferable; 

 but time did not permit modifications of the test equipment before the start of 

 the 1950 fading season. 



