The present invention relates to a microwave oscillator of very high stability.
Signal generation for radar applications, telecommunications and frequency references (such as atomic clocks based on Cs, Rb etc.), and also the developments in digital systems, require oscillators of very high stability that exhibit ever decreasing phase noise.
An oscillator consists of two main components, namely a resonator and an active element or amplifier. The phase noise of oscillators is determined by the combination of the low-frequency noise, the high-frequency noise and the nonlinearities of the active element, the quality factor Q of the resonator and the coupling circuit for coupling between oscillator and active element. The phase noise of oscillators, expressed in the frequency domain, corresponds in the time domain to clock jitter which determines the ultimate precision of all digital signal processing systems and, primarily, analog/digital encoders. The trend in analog and digital systems, in particular the increase in speed and dynamic range of analog/digital converters, is toward a reduction in phase noise of oscillators and the time jitter of clocks.
At the present time, reference oscillators are based on bulk acoustic wave or surface acoustic wave resonators. These resonators are limited, as regards those of highest performance, to frequencies of about 1000 MHz using FBAR (Film Bulk Acoustic-Wave Resonator) technology. Apart from their technological limitations (resonator thickness), acoustic resonators are governed by a fundamental physical law, whereby the product Qf, i.e. the maximum quality factor Q multiplied by the operating frequency f, is a characteristic of the material used. This product may be evaluated by the theory of acoustic loss (an harmonic phonon interactions). Typically, this product is about 10 THz. The quality factor of FBAR-type resonators is thus limited to about 104 for an operating frequency of 1000 GHz.
In practice, it is necessary to generate signals at frequencies very much above 1 GHz. It is therefore necessary to use frequency multipliers. Such an operation degrades the phase noise by at least 20 log N for a multiplication factor N, this resulting from an unavoidable mathematical law.
To achieve the stability performance required for future frequency synthesizers, it will be necessary to use resonators operating at higher frequencies (so as to remove the noise due to the multiplication) and with higher quality factors.
Electromagnetic resonators (metal cavities, dielectric resonators, etc.) allow direct operation at frequencies of several GHz, but their quality factor is limited. For example, for conventional dielectric resonators, a Qf product of 200 THz is obtained, while for sapphire whispering-gallery resonators, at room temperature, a Qf product of 2500 THz is obtained. The phase noise values are thus close to those of multiplied acoustic sources—typically, values of −120 dBc/Hz are obtained at a few kHz from the carrier in the case of the best oscillators.
The limit of current oscillators corresponds to a measurement resolution of analog/digital encoders operating at a frequency of about 1 GHz of 8 encoding bits, which causes a phase jitter of less than 0.3 ps. The systems currently envisaged would require encoders operating at at least 2 GHz with 10-bit resolution, with a phase noise of less than −150 dBc/Hz and with a modulation frequency of 1 kHz. Such performance can be obtained only with resonators having a very high quality factor (Q>106 at 10 GHz for example) combined with oscillator structures that preserve the intrinsic quality of the resonators.
At the present time, the only known solution for increasing the quality factor Q involves oscillators based on cooled electromagnetic resonators. By combining cooled dielectric resonators and superconducting films it is possible to increase the quality factor by two orders of magnitude, i.e., in theory, an increase from 20 to 40 dBc/Hz for the phase noise of the oscillators. However, in practice this improvement is reduced by the sensitivity of the resonators to vibrations and to thermal fluctuations.
The subject of the present invention is a microwave oscillator of very high reference stability, of the resonator type, this resonator exhibiting insignificant sensitivity to vibrations and thermal fluctuations.
The oscillator according to the invention comprises a one-piece dielectric resonator in the form of a right cylinder frustum hollowed out at mid-height along chords of its cross section, so as to leave a central core and two lateral flanges, the drillholes having symmetry of order N, where N≧4, at least the plane faces of the cylinder being covered with a superconducting material, the resonator being placed in a cryogenic chamber and being connected to an amplifier via optimized couplings, and the tuning of the resonator being done by a magnetic field and a phase loop.
The present invention will be more clearly understood on reading the detailed description of several embodiments, given by way of nonlimiting examples and illustrated by the appended drawing in which:
FIGS. 1 to 3 are perspective views of three different embodiments of an oscillator resonator according to the invention;
Since one of the essential components of a microwave oscillator is its resonator, and since the stability of this resonator is affected by mechanical strains, the invention produces it in a different manner from that usually employed. The usual structure of a known resonator generally comprises a cavity in the form of a cylinder frustum closed at its two ends by plane walls made of lanthanum aluminate coated on one face with a single-crystal superconducting material, for example Y1Ba2Cu3O7 containing the actual resonator with its sapphire support and centering foot and two ports for coupling to the cavity. For this purpose, so as to minimize the mechanical strain sensitivity of the resonators, the invention proposes solutions to several undesirable effects in the case of known resonators, these effects being:
The ideal solution would be to be able to deposit the superconducting material on all the faces of the cavity of the resonator. However, as it is impossible for a high-quality film of high-Tc (high critical temperature) superconductor to be grown epitaxially on curved surfaces, the invention proposes to produce a monolithic dielectric resonator, with the general shape of a cylinder frustum, appropriately hollowed out, with direct deposition of the superconducting films on the two plane faces of the cavity, before they are machined. Since the electric field is concentrated on the central core, the degradation of the quality factor by the currents induced in the rest of the structure of the resonator is minimized. The constituent material of the resonator is advantageously single-crystal sapphire.
According to a first embodiment of the invention, shown in
The structure 7 shown in
The structure 10 shown in
The structures of FIGS. 1 to 3 allow the phase noise to be very substantially attenuated between 1 and 10 kHz, at which frequencies there is coincidence between the acoustic wavelengths and the dimensions of the resonator cavity. The values predicted by Leeson's theory (in which the quality factor of a cavity with superconducting films may be estimated by the point at which the colored noise rises above the thermal noise floor) then become achievable. Of course, the constituent materials of these structures must have very low dielectric losses and must be compatible with the deposition of superconducting films.
Next, a theoretical approach allows suitable shapes for the resonators to be determined, for each value of the dielectric constant of the constituent material of the actual resonator, in which shapes the compensation effect occurs, that is to say the diameter/height ratio of the cavity at which the frequency change induced by a slight change in the height is equal and opposite to that induced by the change in diameter resulting from the mechanics equations.
To a first approximation, the resonant frequency of the cavity may be calculated in the configuration described by Hakki and Coleman (see: D. Maystre et al., IEEE MTT, 31, pp 844-848, October 1983) by the equation:
krεJ0(krεr)/j1(krεr)=kr0Z0(kr0r)/Z1(kr0r)
in which:
For perfect compensation, the variation in the dielectric constant ε due to mechanical strains should be taken into account.
To a first approximation, the dependency of the constant ε with respect to the volume V may be deduced from the Clausius-Mossotti equation:
(ε−1)/(ε+2)=NαV
where N is the number of molecules per unit volume and α is the polarizability of a molecule.
The volume change may be calculated from Young's modulus and Poisson's ratio.
In the general case, there is no analytical solution, since:
The resonant frequency is a function of the diameter D of the central core, of the height h and of the permittivity ε.
The frequency fluctuations df related to the dimensions of the resonator cavity are given by:
df=(∂f/∂D)dD+(∂f/∂h)dh.
Values of the parameters (ε, D, h) exist such that, to a first approximation, df=0 for a free structure. By choosing suitable values of the these parameters it is possible to get round the problem of the resonator strains and to obtain a phase noise close to that according to Leeson's theory.
According to a second important feature of the invention, the structure of the oscillator is optimized by trying to fulfill the following requirements:
For these reasons, the invention proposes using the structure of
To ensure that the oscillator is provided with effective and stable cooling, with minimal vibrations, the invention employs a triple chamber 29, as shown diagrammatically in
In the second chamber, the residual gas pressure is slaved in order to precisely control the temperature of the chamber 32 by evaporation and sublimation of the gas, which condenses as a liquid or solid film on the external surface of the case 32. The gas contained in the chamber 32 must remain in the gaseous phase at the temperature within this chamber, so as to ensure temperature uniformity throughout this chamber and to avoid any condensation on the constituents of the oscillator (any condensation would introduce losses into the circuits and would induce a frequency shift). The cooling in the chamber 32 is achieved with minimal vibration, advantageously by a pulsed tube and circulation of the gas (gaseous helium or neon). A thermal bridge is formed between the chamber 32 and the oscillator by means of a flexible metal braid, for example made of copper. The oscillator is suspended inside the case 32 by a suspension system that transmits the minimum possible vibration to it. This suspension system comprises for example, in a manner known per se, suspension cables with non-resonant springs and weights.
Number | Date | Country | Kind |
---|---|---|---|
02/16903 | Dec 2002 | FR | national |
Filing Document | Filing Date | Country | Kind | 371c Date |
---|---|---|---|---|
PCT/EP03/51014 | 12/15/2003 | WO | 6/29/2005 |