Fig. 3: Frequency dependent noise correlations.

a A frequency comb interferogram (black) is dispersively projected onto a charged-coupled device (CCD) array and its correlated intensity noise (blue) modulates the full comb envelope (red). b Upon Fourier transformation, the tomogram’s DC peak is sensitive to the noise, but no change in background signal-to-noise ratio and dynamic range occurs. c In the case of uncorrelated intensity noise between various comb lines, each line of the interferogram exhibits uncorrelated noise, which manifests in the tomogram seen in d as an increase in background signal. e Setup for intensity noise correlation measurement: the source, comprising an external cavity diode laser (ECDL), a semiconductor optical amplifier (SOA), the microchip and an optical spectrum analyser (OSA) generates dissipative Kerr soliton (DKS) and modulation instability (MI) states by laser piezo tuning. Various lines are then filtered from the resulting spectrum using fiber Bragg gratings (FBG), sent through variable optical attenuators (VOA), and sampled with a high-resolution oscilloscope. f, h The cross correlation of lines "1" and "2" for the DKS and MI states, where lines "1" and "2" correspond to 1272 nm and 1320 nm, respectively. The x-axis denotes the relative lag in units of the inverse sampling rate, as derived from the samples per second (Sa s⁻¹), with all color indications shared between sub-figures. g, i Depict the cross power spectral densities corresponding to f and h, respectively. In the spectral plots, the gray shaded region highlight technical noise, likely originating from acoustic modes of the input- and output coupling fibers, while the yellow shaded region indicates the range of typical SD-OCT A-scan rates.