A new method for clock distribution

Abstract

A new alternative is described for the clock-distribution problem. The main feature of the new scheme is that a number of stations arbitrarily located on a single reference line independently derive a clock that is aligned to a common systemwide phase. The central principle is that the midtime between appearances of an isolated pulse traveling down and back on a reference line is the same, regardless of the point of observation. This principle is adapted for clock distribution by the periodic injection of a reference pulse onto a single conductor, reflection of this pulse at the end of the reference line, and generation of a local clock at all stations (such clock being phase-locked to the interval midtime by a special PLL circuit). The paper describes the method and gives a theoretical performance analysis and measured results from an experimental system. The experiments achieve a subnanosecond skew over 30 m on a coaxial cable that has an uncompensated delay of 147 ns. This method should be considered for applications from the millimeter scale on chip to the kilometer scale in radio astronomy antenna arrays. Possible advantages include reduced wiring, reduced tuning, reduced EMI, and/or power consumption and easier in-service growth, relative to conventional methods for the same performance objectives.< <ETX xmlns:mml="http://www.w3.org/1998/Math/MathML" xmlns:xlink="http://www.w3.org/1999/xlink">&gt;</ETX>