A High Precision Standard of Frequency ^ 



By W. A. MARRISON 



Synopsis: A new standard of frequency is described in which three 

 100,000 cycle quartz crystal-controlled oscillators of very high constancy 

 are employed. These are interchecked automatically and continuously with 

 a precision of about one part in one hundred million. They are checked 

 daily in terms of radio time signals by the usual method employing a clock 

 controlled by current maintained at a submultiple of the crystal frequency. 

 Specially shaped crystals are used which have been adjusted to have 

 temperature coefficients less than 0.0001 per cent per degree C. 



TO meet the demands for increased precision in measurement and 

 greater reliability of operation a new reference standard fre- 

 quency system has been developed in the Bell Telephone Laboratories 

 having an absolute accuracy that may be relied upon at all times to 

 better than one part in a million. This reference standard is similar 

 in many respects to one described by J. W. Horton and W. A. Marrison 

 a little over a year ago,^ but a number of important changes have been 

 made which have contributed to increased accuracy and reliability. 



The standard is based on the quartz crystal-controlled oscillator, 

 with a synchronous motor-driven clock, used to determine its rate. 

 It differs from others of the same general type in having a number of 

 similar crystal-controlled oscillators which may be interchanged at 

 will and which are intercompared continuously and automatically 

 with a precision of one part in one hundred million. A number of 

 improvements have been made in the crystal and mounting, and in 

 the circuit, which justify this precision of measurement. 



By far the most important element in a crystal-controlled oscillator 

 is the crystal itself and great care was taken in selecting the type to 

 be used in the new standard. A crystal was required as nearly in- 

 dependent as possible of ordinary variations in temperature and 

 pressure and which could be mounted so as to vibrate freely. The 

 effect of temperature appeared to be especially serious as the changes 

 in frequency thus obtained with an ordinary crystal, even with the 

 best commercial thermal regulators available, are greater than are 

 caused by any other single factor in the new standard. 



It has been known for some time that plates of quartz cut in the 

 plane of the optic and electric axes usually have positive temperature 

 coefficients and that plates cut in the plane of the optic axis but 

 perpendicular to an electric axis have negative coefficients. It has 



1 Presented before Institute of Radio Engineers, April 3, 1929. 

 ^ "Precision Determination of Frequency," by J. W. Horton and W. A. Marrison, 

 Proceedings of Institute of Radio Engineers, Vol. 16, pp. 137-154, Feb. 1928. 



493 



