EVALUATION OF SHORT AND LONG-TERM STABILITY OF THE 2998 MHz REFERENCE-CLOCK TRANSFER SYSTEM P. Lemut 1,2 , B. Batagelj 2,3 , M. Leskovec 2 , J. Tratnik 3 , S. Zorzut 1 1 Instrumentation Technologies d.d., 5250 Solkan, Slovenia 2 COBIK, Velika Pot 22, 5250 Solkan, Slovenia 3 University of Ljubljana, Faculty of Electrical Engineering, 1000 Ljubljana, Slovenia Abstract During the development of the reference-clock transfer system, we encountered several issues, related to the phase drift and phase noise (jitter) measurements. Drift as a phase noise of low frequency, both with random and periodic character, we observe in an offset less than 1 Hz from the carrier. Very few instruments are available for this kind of measurement. We created a prototype of a new instrument – a phase detector that enables long-term drift measurements. A phase detector prototype has been carefully designed and its performance has been evaluated. Besides phase difference measurements, some extended capabilities have been built in to enable full reference-clock transfer system evaluation. On the other hand, commercial-off-the-shelf instruments like Signal Source Analyzers (SSA) are available for the phase noise measurement above 1 Hz from the carrier. INTRODUCTION Large physical experiments like particle accelerator facilities need a reference-clock distribution system for their stable operation. The role of the clock-distribution system is to distribute reference clock to the end user with a constant propagation delay. There are various systems used in the particle accelerator environments. In this article we are focusing on the system where 2998 MHz reference signal is also used for compensation of changes in the optical fiber path. Other systems are mostly based on detection of optical phase. Nevertheless, if the reference signal is available as an electrical signal at the source and at the end point, all the presented methods are applicable. To verify clock-distribution system, performance, development and production testing need to be performed. Phase drift and added jitter need to be measured. Maximum allowed drift and added jitter are both in the range of a few tens of femtoseconds. MEASUREMENT SETUP AND PROCEDURES To confirm proper operation of a reference-clock transfer system a group of measurements of phase and amplitude stability need to be performed. While amplitude stability is rather straight-forward to maintain and measure we will focus on the phase stability. The required stability of 5 fs or roughly 0.005 degrees for the 3 GHz reference clock is difficult to achieve and even more difficult to measure precisely. Absolute value of jitter at the output of the transfer system is important, but users are mostly interested in the added jitter, introduced by the transfer system. Phase noise and its representation as a jitter in the time domain can be divided into a few frequency ranges. Most of the attention is paid to the phase noise in the frequency offsets from 10 Hz to 10 MHz from the carrier frequency. This frequency range can be further divided into several sub-ranges of interest. Usually, within this frequency range most of the spurious signals can be efficiently removed by carefully designing the transfer system. We may assume that only random components like flicker and thermal noise are present. The second frequency range that gives a valuable information of the system stability is for the frequency offset below 1 Hz or even below 0.1 Hz. Phase changes in this range consist of both periodic components due to for example day-night changes or HVAC operation and of random components due to weather changes, solarization, human presence etc. For the transfer system improvements it is of importance to know how the transmission media or the transmitter/receiver electro- optical subsystems are influenced. For the overall performance estimation, only the behavior of the transfer system between the input and output RF connector matters, regardless of the contributions of individual influences. Probably the easiest measurement is the added jitter for the frequency range from 100 Hz to 10 MHz. A Signal Source Analyzer (SSA) can be used for this purpose. If no discrete spectral components are present, we may simply calculate the added jitter from the generator jitter J g and the transfer-system output jitter J out (Fig. 1) as: ! !"" = ! !"# ! − ! ! ! (1) Figure 1: Added jitter measurement. * The operation Centre of Excellence CO BIK is financed by the ERDF and MHEST, Republic of Slovenia Proceedings of BIW2012, Newport News, VA USA MOPG009 Beam Position Monitoring ISBN 978-3-95450-121-2 41 Copyright c ○ 2012 CC-BY-3.0 and by the respective authors