Phase noise measurement of semiconductor optical amplifiers

TL;DR

This work tackles the challenge of measuring phase noise in optical amplifiers by introducing a differential, symmetric Mach-Zehnder interferometer that cancels laser noise and transduces DUT phase fluctuations into an RF beat detectable by a phase-noise analyzer. The key contribution is a validated measurement framework and detailed background-noise modeling that enable true DUT phase-noise assessment at 1.55 μm, yielding an upper bound of $S_\varphi(1\mathrm{Hz})\approx -32\mathrm{dBrad^2/Hz}$ and an Allan deviation of $\sigma_y(\tau)\approx 5.2\times10^{-17}/\tau$. The white phase-noise level scales as $S_\varphi\propto 1/P_i$ and shows no dependence on pump current within the tested range, with measurements reaching $-126\mathrm{dBrad^2/Hz}$ for $f\ge100$ kHz. This approach, applicable to various optical amplifier technologies, has implications for metrology, high-stability laser links, and fiber-optic instrumentation where ultra-low phase noise is essential.

Abstract

We introduce a novel measurement method for the phase noise measurement of optical amplifiers, topologically similar to the Heterodyne Mach-Zehnder Interferometer but governed by different principles, and we report on the measurement of a fibered amplifier at 1.55 $μ\mathrm{m}$ wavelength. The amplifier under test (DUT) is inserted in one arm of a symmetrical Mach-Zehnder interferometer, with an AOM in the other arm. We measure the phase noise of the RF beat detected at the Mach-Zehnder output. The phase noise floor of the amplifier decreases proportionally to the reciprocal of the laser power at the amplifier input, down to $-125$ $\mathrm{dBrad^2/Hz}$ at $f=100$ $\mathrm{kHz}$. The DUT flicker noise cannot be measured because it is lower than the background of the setup. This sets an upper bound of the amplifier noise at $-32$ $\mathrm{dBrad^2/Hz}$ at $f=1$ $\mathrm{Hz}$, which corresponds to a frequency stability of $5.2{\times}10^{-17}/τ$ (Allan deviation), where $τ$ is the integration time. Such noise level is lower than that of most Fabry-Perot cavity-stabilized lasers. These results are of interest in a wide range of applications including metrology, instrumentation, optical communications, or fiber links.

Figures (5)

• Figure 1: Principle of the optical phase noise measurement. AOM: Acousto-Optic Modulator, DUT: Ddevice Uunder Ttest (optical amplifier), PD: quantum PhotoDetectorphotodetector.
• Figure 2: Experimental setup used for optical phase noise measurement. AOM: Acousto-Optic Modulator, DUT: Ddevice Uunder Ttest (semiconductor optical amplifier), EOM: Electro-Optic Modulator, PC: Polarization Controller, VOA: Variable Optical Attenuator, PD: Photodiode, DC: 10 dB Directional Coupler.
• Figure 3: Phase noise, as described in Sec. \ref{['sec:Experiment1']}, measured with $-5$ dBm RF power at the photodiode output. Curve 1: background noise, measured with no DUT, and negligible $\tau_D$. Curve 2: laser noise, measured with no DUT, and $\tau_D=460$ ns. Curve 3: background in the measurement of the laser noise, calculated from curve 1 accounting for $\tau_D=460$ ns. Curve 4: Total noise, DUT plus background, measured with negligible $\tau_D$.Phase noise of the laser, background noise of the setup (with no DUT), and total noise (DUT plus background). The RF power at the photodiode output is $-5$ dBm. The dotted and dash-dot blackgray lines are the phase noise of high-performance cavity-stabilized lasers Guo-2022Xie-2017b. The diamond symbol is the phase noise ($S_{\varphi}(1\:\mathrm{Hz})=-42\:\mathrm{dBrad^2/Hz}$) of the laser stabilized to a cryogenic silicon cavity Kedar-2023. Such phase noise level is calculated from $S_\mathsf{y}(1\:\mathrm{Hz})=1.5{\times}10^{-33}~\mathrm{Hz^{-1}}$Kedar-2023 using $S_\varphi(f)=(\nu_0^2/f^2)\,S_\mathsf{y}(f)$.
• Figure 4: White phase noise of the DUT measured at $I_P=350~\mathrm{mA}$ pump power, for different values of the optical power $P_i$ at the DUT input. From the right hand to the left, the RF power is reduced by detuning the attenuator at the photodetector input. The background noise is shown for comparison.
• Figure 5: Phase noise floor of the optical amplifier versus the SOA input optical power $P_i$, for several values (600, 350 and 150 mA) of the SOA driving current $I$. Only 1 point was taken for $I$ = 150 mA. In this case, the output RF signal was too low for lower values of $P_i$.