Abstract:We present an investigation of the white noise plateau in residual phase noise measurement of an optical frequency comb for short and long external laser cavity lengths.OCIS codes: (140.5960) Semiconductor Lasers; (140.4050) Mode-locked lasers
MotivationSemiconductor harmonically mode-locked lasers have a number of features which make them attractive for use in communications, sampling, and metrology applications [1,2] For some of these applications, low timing jitter and narrow optical comb linewidth is required for good sampling resolution or increased stability.In semiconductor phase noise performance plots, we generally distinct characteristics at low, middling, and high offset frequencies relative to the carrier [3,4]. For good phase noise performance at high offset frequencies, we can implement a high finesse intra-cavity to suppress competing longitudinal mode-groups [5], and we can lower the shot noise floor by increasing photodetected optical power. In this submission, we address phase noise performance at middling offset frequencies corresponding to the white noise plateau. By decreasing optical linewidth, the corner, or "knee" frequency of the white noise plateau should also decrease proportionately [6], and we aim to hit the shot noise floor at lower offset frequencies and thereby reduce timing jitter.
BackgroundIt is well known that laser cavity length determines longitudinal mode linewidth [7]. It is also known that a mode-locked pulse train's temporal coherence is inversely related to longitudinal mode linewidth [8]. For pulses separated in time by more than the coherence time, we expect to see no coherence, and, importantly, no noise correlation. Thus the noise contribution at low offset frequencies is dominated by uncorrelated white noise. This result has been shown in a previous publication without the use of the etalon to suppress supermode contributions to phase noise [6], and here we will attempt to implement the concept into our existing laser architecture [9].
Setup and DataThe laser's setup is shown in Fig. 1 and a similar architecture is discussed in [9]. Here we perform measurements at the shortest manageable cavity length (22m, f0 = 9MHz) and for a longer cavity length (351 m, f0 = 570 kHz). Fig. 2 shows the change in knee location for the residual phase noise measurement when bypassing the etalon in the laser (not PDH-locked). Fig. 3 shows the residual phase noise measurement for the short cavity. As of the time of this submission, the PDH lock on the longer cavity is limited due to the higher required dynamic range for the fiber stretcher, but the dashed line in Fig. 3 shows where we expect to measure the new knee location once we have ability to stabilize the laser. For equal shot noise floor and noise power at 1 Hz offset, we expect a reduction in timing jitter between 10% and 20%. This represents a promising improvement in the timing jitter of our optical frequency comb sources.
References[1] P.