APPLICATION OF PRINTING TELEGRAPH 



611 



the per cent of errors corresponding with the average noise condition is 

 a much more significant figure than the average per cent of errors. For 

 example, in the Rochester tests Fig. 6 indicates that the per cent of er- 

 ror corresponding to average noise condition is 0.28 per cent while the 

 observed average of the daily per cents of errors is 6.44 per cent. It is 

 more useful to know that half of the time the copy will be better than 



5z: 



a\ 



\ • 



^f 01 80 



O o 



xl^o 



5q:<60 



o 



<ot 



:P;; 



O 10 20 30 40 



HOULTON NOISE READINGS ADJUSTED 



TO ROCHESTER SIGNAL-TO-NOISE RATIO 



(db)AT V.F. FILTER OUTPUT 



(T in^o 40 



UJ uj5:CL 

 Oui 



UJ 'i^ca 



> O<oa: 



H l-<0 20 



< ZgO 



D Og 



D UJ (r 



O Q- UJ 



0.4 0.8 1.2 1.6 



PER CENT OF ERRORS ON ROCHESTER 

 PRINTER COPY 



pjg 6 — ^Cumulative curves of signal-to-noise ratio derived from Houlton noise 

 observations and of per cent of errors on the printer copy observed at Rochester, 



0.28 per cent and half of the time worse, than to be unduly influenced 

 by the effect on the average per cent of error of a few days in which the 

 copy is almost all errors. 



The results of the Rochester tests may be briefly summarized by 

 giving a few figures which are based on the data obtained. A five-kilo- 

 watt station on long waves with a reasonable antenna, say 20 per cent 

 efficient, would radiate one kilowatt. Assume that the local noise con- 

 ditions are the same at the receiving station as those which have been 

 used for the 9:00 p.m. values in Fig. 7 for Rochester, N. Y., variations. 

 (These are obtained by applying a correction factor to the Houlton, 

 Maine, noise observations for 1930.) Then the per cents of errors in the 

 teletypewriter copy during the evening periods at different distances * 



* As the distance varies between transmitter and receiver with the radiated power 

 a constant, there is a variation in received signal field. If the noise is assumed to be 

 fixed, this variation in distance will result in a variation in signal-to-noise ratio. 

 Many of the commonly used radio transmission formulas take the form: « 



,-300 X We-"Di\' 



E = y'P—D — 



For these calculations we have assumed x = 1.25 and from the field strength measure- 

 ments at Rochester a = 0.023. P is measured in kilowatts radiated, D and X in 

 kilometers, and E in microvolts per meter. , , . r u 



The various signal-to-noise ratios can then be translated into rates ot error by 

 use of Fig. 5. 



