Fig. 2: Analysis results of the technical and squeezed noises under different conditions.

a Dependence of the out-of-loop technical noise \({{\rm{T}}{\rm{N}}}_{{\rm{O}}{\rm{O}}{\rm{L}}}\) on the analysis frequency. Experimental parameters are listed as: BS splitting ratio: 99:1; feedback bandwidth: 2 MHz; free-running noise: −125 dB/Hz; detectable in-loop power: 10 mW; photodetector electronic noise is −170 dB/Hz; squeezing strength: −10.5 dB. The final theoretical limit is simulated by considering a 99 mW optical power detected by the il-PD without electronic noise (trace V). In practice, the detectable power is restricted to the 10 mW saturation power of the il-PD with non-negligible electronic noise (traces I–IV). The dashed line is calculated by the factor of \({f}_{{\rm{B}}}/f\). b Dependence of the squeezing strength of bright squeezed light at analysis frequency of 1 MHz on the output power, which represents the maximum measurement frequency in the squeezing bandwidth. All these results are inferred from a and Eqs. (1)–(3), and are normalized to SNL