Observatoire des Sciences de l’Univers Terre Homme Environnement Temps Astronomie

Accueil > Projets et réseaux


par François Vernotte - publié le

In a nutshell

The Oscillator-IMP project targets at being the world-leader facility dedicated to the measurement of noise and short-term stability of oscillators and devices in the whole radio spectrum (from MHz to THz), including microwave photonics, widely available to Agencies, to research institutions and to private companies in the spirit of global competition and economy. The scope spans from routine measurements to the research on new oscillators, components, and measurement methods.

Already listed in the Calibration and Measurement Capabilities (CMCs) published in the BIPM Key Comparison Database (KCDB) as the French reference laboratory for short-term stability and for phase noise associated to the LNE, we aim at a significant step forward in sensitivity, accuracy and range, and also at upgrading the COFRAC accreditation (National accreditation). Finally, this project will work in tight collaboration with the FIRST-TF LABEX network (starting fall 2011).

Oscillators and short-term stability

Time, and equivalently frequency, is the most precisely measured physical quantity. Accuracy ranges from parts in 10–5 (watch) to <10–15 (primary laboratories), and stability is even higher. Therefore, most domains of science and technology rely on time-and-frequency (T&F) metrology, and in turn on stable oscillators.

Frequency stability is expressed in terms of Allan deviation σy(τ) of the fractional frequency y, as a function of the measurement time τ. For fast phenomena (τ ≤ 10–2…1 s), the power spectral density Sφ(f) of the random phase φ(t) is more suitable than the time-domain measurement. For historical reasons, engineers prefer the practical quantity L(f) = ½ Sφ(f). The terms “spectral purity” and “phase noise” are often encountered.

Scanning the technology, we notice that virtually all systems need that a master oscillator be stable for an appropriate τ, while the accuracy – when needed – ultimately relies on a primary standard. This pattern is found in the Galileo pendulum (τ = 105…106 s), steered to the rotation of the Earth, in radars (τ = 10–6…10–2 s), in telecommunication systems (τ = 1…105 s, depending on the oscillator role), in computer boards (τ < ≈1 µs), in particle accelerators (τ ≤ ≈100 ms), in very-large baseline interferometry (τ = 10–1…104 s), and in space missions (τ = 1…103 s), to mention some. Even satellite navigation relies on small onboard atomic clocks stable for τ ≤ 106 s (≈2 weeks), controlled to primary Cs fountains. In all the cases mentioned, the short-term stability, i.e. the stability for small τ, is the most desired feature. Needless to say, high short-term stability and low phase noise are of paramount importance in space and military electronics, where France is the European leader.

The international context

Top-level T&F metrology is carried on by the National laboratories and coordinated by the BIPM. The BIPM is in charge of the international time scale TAI. About 70 laboratories maintain a national time scale and contribute to TAI. The National labs are mainly concerned by primary frequency standards, and give comparatively smaller importance to the short-term stability. The assessed metrological performance of the laboratories is published in the BIPM KCDB as a part of the Mutual Recognition Arrangement of the International Committee for Weight and Measures (CIPM MRA). In contrast with the relatively large community of timekeeping, only 8 laboratories have recognized capability in the measurement of short-term stability, and 1 has also a recognized capability of phase-noise metrology. The LNE (service provided by us under the name LNE-LTFB) is the one and only laboratory in the BIPM KCDB for both. The NIST is likely the world leader, even if there are not CMCs declared in the KCDB. However, the NIST is clearly oriented to the American economy.

In the past, the knowhow developed to test the primary frequency standard were suitable to most applications demanding short-term stability. With the progress of technology this is no longer true. Now spectral purity and short-term stability constitute an emerging branch of T&F metrology (for short, we may write “short-term stability” referring to both these sub-domains). Moreover, users are not sufficiently aware that their problems fall there, and that know very little about the help that they can find. This has been observed even in high-rank labs, like CEA/Grenoble, ANL/Argonne and DESY/Hamburg. Since we have high profile in the domain, these laboratories searched for our help with short-term stability and phase noise. Dr. Rubiola also visited the NASA JPL to help with phase noise, and lectured at OEwaves, a company that produces oscillator technology mainly for military applications. CERN is now interested in sapphire oscillators, and MPI-QO (München) already asked to borrow one from us.

As a conclusion, the National laboratories target timekeeping and fundamental science, while the private companies seem to be driven by specific problems. With the remarkable exceptions of NIST and of some sophisticated optical cavities for atomic standards, the primary laboratories never really joined in this business. In the future, the short-term stability will be clearly identified as a separate branch and will have its own role in technology, though always close to fundamental T&F metrology. This is an opportunity with still no competitors in Europe.

The Oscillator-IMP project

The platform consists of highly sophisticated instruments for the measurement of short-term stability and spectral purity, and of the most stable oscillators.

  • High stability reference oscillators
    • Three Cryogenic Sapphire Oscillators, each provides stability of 2×10–15 at τ = 1…105 s. One of these oscillators is transportable, and a van adapted to this purpose is available.
    • Three Hydrogen masers, each provides stability of 2×10–15 at τ > 103 s.
    • Two femtosecond fiber-laser combs for wide-range frequency synthesis, stabilized to a ULE Fabry-Pérot etalon for σy(τ) < 10–15 at τ ≤ 1 s
    • Distribution of the reference signals over the campus
    • Transportable quartz-oscillator ensemble for off-site tests, targets 50 kg mass including battery for 48H autonomy, and 10–14 stability measurements for τ = 1…100 s using statistics.
  • Specialized instrumentation
    • Instruments for the measurement of phase noise, amplitude noise, and Allan variance
    • Three-corner-hat statistical measurements (Y-∆ scheme), in both time and frequency domain. This improves the limit by 6–20 dB (depending on τ) vs. the stability of the available references. The characterization of oscillators of stability at background level of parts in 10–16 is expected.
  • General equipment for microwaves (up to 200 GHz) and microwave photonics.
  • Basic equipment for stability measurements in perturbed-environment conditions
    • Shielded chamber and related equipment for electromagnetic compatibility tests.
    • Vacuum chamber, vibrating plate, programmable-temperature chamber, etc.
  • The French national time scale, coordinated at higher level, is made available by us
    • Three Cs-beam frequency standard in controlled environment
    • TWSFTFT, differential GPS, and carrier-phase GPS synchronization
    • Our H-maser and sapphire oscillators steered to the time scale
  • Approval of CMCs for publication in the BIPM KCBD, COFRAC accreditation, and ISO-9000 certification, depending on the type of test.

The platform is intended to measure frequency stability up to 106 s integration time. However we agree that “short-term” should be τ ≤ 105 s (≈1 day), we allow one-decade overlapping to the timekeeping community as a tribute to high measurement consistency.

Current status and timeline

Approximately ⅓ of the equipment listed is available, now or by the end 2011, and funded separately. This is the case of the 3 Cs standards, the 3 cryogenic oscillators, and one ULE-FP-stabilized laser comb. Completing will take three years, though most of the features will be available in less than two years.