Application of the optical fiber to generation and measurement of low phase noise microwaves

Kirill Volyanski (1, 2) , Johann Cussey (1) , Hervé Tavernier (1) , Patrice Salzenstein (1) , Gérard Sauvage (3) , Laurent Larger (1) , Enrico Rubiola (1)

1 FEMTO-ST - Franche-Comté Électronique Mécanique, Thermique et Optique - Sciences et Technologies (UMR 6174)
2 GUAP - State University of Aerospace Instrumentation
3 AEROFLEX - AEROFLEX Test Solutions

The optical delay line proved to be a method to measure the phase noise of microwave oscillators with high sensitivity. The delay, inherently, turns frequency fluctuations into phase fluctuations. Hence, a mixer is used to compare the phase of the oscillator signal to a delayed copy, from which we measure the oscillator phase noise. This article reports on the progress in this type of instruments in our laboratory. For practical reasons, the delay is implemented with an optical-fiber channel, either at 1.3 or 1.55 µm wavelength, modulated in intensity. The laser Relative Intensity Noise (RIN) turns out to be a critical parameter because it converts the RIN into near DC noise through the mixer offset sensitivity to power. The best semiconductor lasers we can find show a RIN of about -155 dB/Hz. Additionally, in our experience the simple microwave photodiodes are to be preferred to the photodiodes with integrated transconductance amplifier because of the lower noise. This seems to be a technical issue, rather than a general property. Thus, we used a separate amplifier, based on SiGe technology for lowest flicker. The optical fiber is temperature stabilized by an Aluminum mass and a sub-milliKelvin electronic control. Other components, like the Mach Zehnder intensity modulator, seem to be less critical for noise. A single channel version was realized and tested with a microwave synthesizer and with a sapphire whispering gallery oscillator at 10 GHz carrier frequency. Using a 2 km delay line (Τ=10 µs) the measured 1/f3 noise is b-3 = -12 dB.rad²/Hz, which matches the a-priori knowledge of the oscillator 1/f3 noise. This means that the instrument sensitivity is higher than this value, thus it is sufficient to measure a room-temperature sapphire oscillator without need of correlation. At higher Fourier frequencies, the instrument background noise is of -145 dBrad²/Hz at 10 kHz off the carrier. Unfortunately, with other source types (eg, a synthesizer) the background noise can be higher because of the effect of am noise. Another test of the noise floor consists of shortening the optical delay to a negligible value. In this way, the oscillator noise is rejected. Unfortunately, this is only a qualitative test because it hides the noise of the optical fiber, due to Rayleigh scattering and other optical phenomena. The background noise is reduced proportionally to 1/√m, where m is the number of averaged spectra, by correlation and averaging on two fully independent channels that measure the same oscillator. Using 2 km optical fibers and averaging on 200 spectra, the background noise is of -110 dBrad²/Hz at 100 Hz off the carrier, and of -160 dBrad²/Hz at 10 kHz. Re-using the parts of the two-channel system, we assemble single-channel system with matched 10 µs optical delays at the two inputs of the mixer, which rejects the noise of the oscillator. The 1/f noise measured in this condition, b-1 = -113 dBrad²/Hz referred to one channel, is the background noise of the system without correlation, which includes amplifiers, detector and optical fiber.

Domaines: Optique / photonique; Electromagnétisme; Electronique

Kirill Volyanski, Johann Cussey, Hervé Tavernier, Patrice Salzenstein, Gérard Sauvage, et al.. Application of the optical fiber to generation and measurement of low phase noise microwaves. 22nd European Frequency and Time Forum, Apr 2008, Toulouse, France. pp.NA. ⟨hal-00277058⟩