Higher Timing Accuracy, Lower Cost
November 1, 2009 By: Aiden Morrison GPS WorldMillisecond-Level Oscillator Phase-Noise Estimation
Aurora Borealis seen from Churchill, Manitoba, Canada. Ionospheric scintillation
research can benefit from this new method.
Two broad user groups will find important consequences in this article:
Time synchronization and test equipment manufacturers, whose GPS-disciplined oscillators have excellent long-term performance but short- to medium-term behavior limited by the quality, and therefore cost, of the integrated quartz device. This article portends a family of devices delivering oven-controlled crystal oscillator (OCXO) performance down to the 10-millisecond level, with an oscillator costing pennies, rather than tens or hundreds of dollars. Applications include ionospheric scintillation research (above).
High-performance receiver manufacturers who design products for high-dynamic or high-vibration environments (see cover) where the contribution of phase noise from the local oscillator to velocity error cannot be ignored. In these areas, the strategy outlined here would produce equipment that can perform to higher specifications with the same or a lower-cost oscillator.
The trade-off requires two tracking channels per satellite signal, but this should not pose a problem. At ION GNSS 2009, manufacturers showed receivers with 226 tracking channels. There are currently only 75 live signals in the sky, including all of GPSL1/L2/L5 and GLONASS L1/L2. — Gérard Lachapelle
If the channel data within a GNSS receiver is handled in an effective manner, it is possible to form meaningful estimates of the local-oscillator phase deviations on timescales of 10 milliseconds (ms) or less. Moreover, if certain criteria are met, these estimates will be available with related uncertainties similar to the deviations produced by a typical oven-controlled crystal oscillator (OCXO). The processing delay required to form this estimate is limited to between 10 and 20 ms. In short, it becomes possible in near-real-time to remove the majority of the phase noise of a local oscillator that possesses short-term instability worse than an OCXO, using standalone GNSS. This represents both a new method to accurately determine the Allan deviation of a local oscillator at time scales previously impractical to assess using a conventional GNSS receiver, and the potential for the reduction in observable Doppler uncertainty at the output of the receiver, as well as ionospheric scintillation detection not reliant on an expensive local OCXO.
Concept. Inside a typical GNSS receiver, the estimate of the error in the local oscillator is formed as a component of the navigation solution, which is in turn based on the output of each satellite-tracking channel propagating its estimate of carrier and code measurements to a common future point. While this method of ensuring simultaneous measurements is necessary, it regrettably limits the resolution with which the noise of the local oscillator can be quantified, due to the scaling of non-simultaneous samples of local oscillator noise through the measurement propagation process. To bypass these shortcomings requires a method of coherently gathering information about the phase change in the local oscillator across all available satellite signals: to use the same samples simultaneously for all satellites in view to estimate the center-point phase error common across the visible constellation.
To explain how this is feasible, we must first understand the limitations imposed by the conventional receiver architecture, with respect to accurately estimating short-term oscillator behavior, and subsequently to determine the potential pitfalls of the proposed modifications, including processing delays needed for bit wipe-off, expected observation noise, and user dynamics effects.
1 2 3 4 5 6
RSS
Become a Fan
Join our Network
Follow Us
Please send any technical comments or questions to our webmaster.