Digital Clocks for Synchronization and Communications

When clock specifications are discussed, noise analysis of clock signals is inevitable. Parameters described in Chapter 13 are used to specify those clock characteristics that are influenced by noise. While practical data can be measured with commercial measuring equipment, some understanding of the physical mechanism in the measurement process is required to get accurate data and to process it. This chapter explains the measurement mechanisms and shows the connection between variance in time domain and power spectrum in frequency domain.
Frequency counters directly measure the signal frequency from which the two-sample variance can be calculated. A frequency counter measures, however, the mean frequency within its gate time from t i to t i + ? and cannot record the instantaneous value. Naturally, ? corresponds to the time over which the frequency averaged. Therefore, the normalized frequency deviation can be expressed as
where f o is the nominal frequency of the signal to be measured and f i is the actual measured value. The two-sample variance can be calculated by repeating the frequency measurement M times, based on (14.2). M 1 differential data is obtained by these measurements.
Here the number of measured samples is finite and is limited to M 1 different values, and thus, (14.2) differs from the two-sample variance defined in (13.25). In practice, of course, we cannot help but...