Fig. 1. Schematic of the self-injection locked laser via the HCF-FP. A compact polystyrene box is filled with gel-packs that have a large thermal capacity, passively reducing temperature fluctuations.

Fig. 2. (a) Simulated frequency noise spectrum when considering self-injection locking noise and thermal noise. (b) Lorentzian linewidth as a function of the HCF-FP length.

Fig. 3. (a) Schematic of the fabricated HCF-FP. Insert shows a microscope image of the used HCF (DNANF). (b) The set-up to measure the transmission signals of the HCF-FP. PC, polarization controller; PD, photodetector.

Fig. 4. Transmitted signal through the HCF-FP measured in the set-up shown in Fig. 2(b). (a) Transmission over 50 MHz; (b) detail of transmission of single peak.

Fig. 5. (a) Experimental set-up for long-term frequency error measurement; (b) optical frequency discriminator set-up for frequency noise measurement. PC, polarization controller; PD, photodetector.

Fig. 6. (a) Frequency error as a function of time achieved here and its comparison with the stability of a laser locked to an SMF delay line placed in a vacuum [29]. (b) Allan deviation calculated from the data shown in (a) and its comparison to measurements reported by manufacturers of commercially available narrow-linewidth lasers [3032" target="_self" style="display: inline;">–32], including OEwaves, which are also based on self-injection locking.

Fig. 7. Measured frequency noise via strong self-injection ratio of −15  dB, together with free-running laser and measurement noise floor.

Fig. 8. Measured frequency noise with different self-injection ratios.

Fig. 9. Comparison of the measured and simulated frequency noise for injection ratio of −15  dB (green) and −30  dB (red). Dashed gray lines show the thermo-conductive noise (which is the same for both cases) and noise contribution due to the self-injection locking.

Fig. 10. RIN of the free running and SIL lasers. Shot noise expected for 0 dBm power output is also shown.