Jacques Groslambert

Flicker noise measurement of HF quartz resonators

Frequency flicker of quartz resonators can be derived from the measurement of S/sub /spl phi// (f), i.e., the power spectrum density of phase fluctuations /spl phi/. The interferometric method appears to be the best choice to measure the phase fluctuations of the quartz resonators because of its high sensitivity in the low power conditions, which is required for this type of resonator. Combining these two ideas, we built an instrument suitable to measure the frequency flicker floor of the quartz resonators, and we measured the stability of some 10-MHB high performance resonators as a function of the dissipated power. The stability limit of our instrument, described in terms of Allan deviation /spl sigma//sub y/(/spl tau/), is of some 10/sup -14/.

Oscillator noise analysis: multivariance measurement

Since the noise altering the output signal of oscillators may be modeled as power laws in the spectral density of frequency deviation, oscillator noise analysis is the measurement of the level of each power law noise. The principle of this new multivariance method consists of obtaining the noise-type contributions with different variances and different integration time values. All the data obtained from the different variances with the different integration times are then operated simultaneously. Thus, the most probable measurement, in the sense of least squares, is obtained for each type of noise. This method lends itself to an estimation of the uncertainty of the noise-type contribution measurement, taking into account the dispersion of the variance results. >

Phase noise in the regenerative frequency dividers

The authors report on a theoretical analysis of the phase noise in regenerative dividers, and provide design rules for the best spectral unity. The spectral purity is, in fact, the reason why regenerative dividers may be preferred to simpler schemes whenever that characteristic is a fundamental requirement. This class of dividers is suitable for high frequencies, out of reach to other techniques. The nucleus of the theory is the description of the mixer, driven by coherent signals, in terms of differential phase gain, so as to relate noise at the divider output to the noise generated inside the divider itself. After introducing a model of the mixer and some related measurement techniques, it is shown that the most favorable noise condition for the divider is reached by tuning it off the maximum output amplitude. Methods for approximating this condition are then outlined. The experiments prove the feasibility of the proposed approach, and reported results show a phase noise reduction of 10 dB with respect to the maximum amplitude condition. >

Very high frequency and microwave interferometric phase and amplitude noise measurements

The interferometric technique allows close-to-the-carrier measurements of both phase and amplitude noise, improving the instrument noise floor by 10–25 dB as compared to the traditional method based on a saturated mixer. Principles and basic equations describing the noise measurement system are given, together with design strategies suitable to microwave and very high frequency bands. Two prototypes, operating at 9 GHz and 100 MHz are discussed in detail. The relevant features of these prototypes are the capability to operate in a wide power range, below 0 dBm and above 20 dBm, and low noise floor. The latter is about −180 dB rad2/Hz (white) and 150 dB rad2/Hz (flicker) at 1 Hz Fourier frequency, at carrier power from 9 to 15 dBm.

A new 'filtered Allan variance' and its application to the identification of phase and frequency noise sources

Digital noise generators are described as is their use to generate sample series that simulate frequency samples as given by a counter. Applying the Allan variance to the samples of six generators (white phase, flicker phase, white frequency, flicker frequency, frequency random walk, filtered flicker frequency) yields the expected theoretical slopes. The method used consists in filtering the f/sup -1/ and f/sup 0/ phase noise by means of a digital filter in the time domain, (using the Z transform), which yields f/sup -3/ and f/sup -2/ noises, which thus can be identified by Allan variances. The combination of the digital filter and the Allan variance corresponds to a novel variance, which is described and compared to the well-known modified Allan variance.<<ETX>>

A new method of measurement of the different types of noise altering the output signal of oscillators

Oscillator noise is generally modeled by a power law spectral density. Thus it is possible to characterize different noise sources, each of them corresponding to a particular power law. The measurement of the contribution of these sources is necessary to know their origin and to remedy these causes in order to improve oscillator performance. Usually, an estimation of the different types of noise present in a signal is obtained by using a variance. However, the sensitivity of these variances differs for each type of noise and then limits this method. On the other hand, the use of several variances, each of them more sensitive to one type of noise, permits one to notably improve the measurement accuracy. The method suggested here uses as many different variances as there are types of noise to measure. The improvement of measurement accuracy of the noise coefficient is discussed in this paper. >

High spectral purity frequency sources using low noise regenerative frequency dividers

Regenerative frequency dividers with input frequency in the range of 100 MHz to 1 GHz were studied. The phase noise of such dividers shows a 1/f spectrum corresponding to 155 dB. These performances are much better than those of GaAs digital frequency dividers. Using these regenerative frequency dividers at the VHF output of a synthesizer, a frequency source of high spectral purity was achieved, working in the range of 5 to 20 MHz. It can be used to test the spectral purity of oscillators in this frequency range.<<ETX>>

A new spectrum analyzer using both analog and digital filtering via Hadamard variance

In this paper, the author shows how to simplify the complex response of the Hadamard variance by analog filtering to design a spectrum analyzer. The realized spectrum analyzer is described.

Spectral and Short Term Stability Measurements

The noise performance of an oscillator can be given either in the spectral or in the time domain. Two types of apparatus are generally necessary to measure these noise characteristics, spectral analyzers and frequency counters. The system described uses spectral density to time domain conversion and measures both the short term frequency stability and the phase spectral density of an oscillator. Bias functions, depending on the spectral density, are calculated. They are used to determine systematic errors introduced by the apparatus.

Automatic Plotting Systems for Frequency Stability Curves and Discrete Spectral Density of Frequency Standards

In this paper, the authors describe two automatic systems to analyze the performance of quartz oscillators and especially frequency standards. The first system is the I(?) plotter used to the frequency stability measurement of oscillators in time domain. The second system draws the discrete spectrum of an oscillator. A computer coupled to the analyzer allows us to do a statistical analysis on the measurement.