香港天文學會-掩星組 - Powered by Discuz! Board

標題: 使用 GPS 接收器作為 NIST 計時朔源參考 [打印本頁]

作者: occult 時間: 2009-9-10 21:36 標題: 使用 GPS 接收器作為 NIST 計時朔源參考

原文有關使用 GPS 計時朔源文件網頁
http://tf.nist.gov/service/gpscal.htm

GPS receivers often provide a 1 pulse per second (pps) timing output,as well as standard output frequencies such as 5 and 10 MHz. If properly designed and used, a GPS receiver can provide traceability tothe NIST frequency standard. The NIST GPS data archiveshows the frequency uncertainty of the signals broadcast from each GPS satellite with respect to NIST. However, before claiming NISTtraceability using GPS, please read the following sections.

作者: occult 時間: 2009-9-10 21:36

朔源的重要性
甚麼叫做朔源 (traceability)? 按照 ISO《International Vocabulary of Basic and General Terms in Metrology》文件，簡稱 (VIM) [1], 朔源定義如下：

The property of a result of a measurement or the valueof a standard whereby it can be related to stated references, usually national or international standards, through an unbroken chain ofcomparisons all having stated uncertainties.

Since many companies now seek registration or compliancewith ISO-9000 quality standards, more and more importance is beingplaced on traceability. ISO Guide 17025 [2] is the internationallyrecognized document that lists the requirements for competence ofcalibration and testing laboratories. Section 5.6.21 of Guide 17025states:

For calibration laboratories, the program forcalibration of equipment shall be designed and operated so as to ensurethat calibrations and measurements made by the laboratory are traceableto the SI (Systeme International) units of measurement. Traceability ofmeasurement shall be assured by the use of calibration services fromlaboratories that can demonstrate competence, measurement capability,and traceability. The calibration certificates issued by theselaboratories shall show there is a link to a primary standard or to anatural constant realizing the SI unit by an unbroken chain ofcalibrations.

In short, in order to meet Guide 17025 requirements,calibration and testing laboratories must demonstrate that theircalibrations are traceable to national standards. In many cases, UnitedStates laboratories must show traceability to NIST. Since manylaboratories use GPS receivers as a frequency reference and since GPSis not a NIST generated service, this raises the following questions:"Is GPS a NIST-traceable frequency reference? If so, what is theuncertainty?" The answers to these questions are discussed below

作者: occult 時間: 2009-9-10 21:37

溯源鏈
The definition of traceability tells us that a traceable measurementrequires an "unbroken chain of comparisons all having stateduncertainties." In order to show traceability to NIST through the useof GPS, the unbroken chain must stretch from the measurement made withGPS back to NIST.
The time and frequency reference for GPS is provided by the UnitedStates Naval Observatory (USNO). The NIST and USNO frequency standardsare regularly compared and equivalent at their point of origin to 1 x10-13 or less. [3] One way to establish the traceability chain is toshow the uncertainty of the GPS constellation relative to the USNOfrequency standard, and then show the uncertainty of the USNO frequencystandard relative to the NIST frequency standard. Raw data from thefollowing sites can be used to compute the uncertainty:

Asecond, more convenient way to establish the traceability chain isthrough a direct comparison of GPS to the NIST frequency standard. Thiscomparison is made through NIST monitoring of the GPS satellites asdiscussed in the next section

作者: occult 時間: 2009-9-10 21:37

NIST 監察 GPS 廣播訊號
In order for continuous traceability to be established, continuouscomparisons must be made. In a presentation at the 5th U. S.-ItalyBilateral Seminar, the Acting Deputy Directory of NIST expanded on thedefinition of traceability:

It is noted that traceability only exists whenscientifically rigorous evidence is collected on a continuing basisshowing that the measurement is producing documented results for whichthe total measurement uncertainty is quantified.

Using these comments as a guideline, Ehrlich and Rasberry [4] of NIST state that:

A single measurement result is sufficient to establishuncertainty relationships only over a limited time interval, and thatdirect periodic comparisons are otherwise required.

For this reason, NIST compares the frequency recoveredfrom GPS to the national frequency standard 24 hours per day toestablish continuous traceability. The signal broadcast from eachsatellite is monitored for the entire time that the satellite isvisible from the NIST laboratories in Boulder, Colorado. The results ofthese comparisons are published (updated daily) in the NIST GPS data archive.
The data archive lists daily time and frequency offsets for each GPSsatellite. These data for the previous UTC day are made available eachmorning at about 1600 UTC. The archived data are obtained by comparinga "typical" GPS receiver to the NIST frequency standard. These data canbe used to support claims of frequency traceability to NIST through theuse of GPS signals.
Keep in mind that NIST monitors the satellites only while they arevisible from Boulder, Colorado. Data broadcast by the satellites whenthey are not visible to NIST are not published in the data archive.Using more than one monitoring station is a future possibility.

作者: occult 時間: 2009-9-10 21:37

是否所有 GPS 接收器都能夠朔源標準時間?
Even though the GPS broadcasts are continuously monitored by NIST, not all GPS receivers are suitable for use as traceable frequencystandards. Remember that the definition of traceability states thattraceability is the property of "the result of a measurement." Theuncertainty relative to NIST can vary widely depending upon the GPSreceiver used to perform the measurement. The uncertainty is alsodependent upon the method or procedure used to perform the measurement.
To explain why different GPS receivers behave differently requiresdiscussing the similarities and differences of the commerciallyavailable models. Hundreds of companies sell GPS products, and at leasta dozen manufacturers advertise their units as time and frequencystandards. Most receivers are designed using an OEM GPS "engine." Theseengines are small circuit boards or chipsets that require a powersupply, antenna, and control software to use. They typically have acomputer interface and provide a 1 pps output.
Commercial GPS receivers share several characteristics:

Most GPS receivers automatically select the satellites used inthe timing solution. This makes them easy to use. Often, you simplyturn the receiver on and wait for a signal to be acquired. However,different algorithms are used to select satellites, and each receiverhas its own thresholds at which it decides to keep, drop, or acquire asatellite. Some algorithms choose the satellites that provide the bestgeometric dilution of precision (GDOP). Others choose the satelliteshighest in the sky after a fixed position has been entered. Somealgorithms limit the timing solution to just one, or just a fewsatellites. Others can use as many as 12 satellites in the solution.For this reason, two GPS receivers can obtain very different resultseven when connected to the same antenna in the same location.
MostGPS receivers have poor short term stability. The models with the bestshort term stability typically discipline an oven controlled quartzoscillator (OCXO) or a rubidium oscillator to the GPS signal. However,many receivers do not discipline an oscillator at all. Instead, theydivide the output of a small temperature controlled crystal oscillator(TCXO) to 1 pps, and then synchronize the 1 pps to the GPS signal. TheTCXO free runs and the receiver accumulates time errors until the totaltime error approaches a threshold (a multiple of the half period of theoscillator), and then generates a phase step that reduces the timeerror to a minimum. Some receivers step phase in increments of 100 nsor less, but some use increments of 1 msor larger. If the TCXO is offset in frequency by 1 x 10-7 (typical), a100 ns phase correction is needed every second. As a result, the shortterm stability of these models is very poor, but their long termperformance may be equivalent to models that discipline a quartz orrubidium.
Some GPS receivers are suitable for timingapplications, but are not suitable as a frequency reference. Forexample, a receiver that produces a 1 pps output for timingapplications, might do a poor job of producing standard frequencieslike 5 and 10 MHz. In some cases, the 1 pps output is not in phase withthe standard frequency outputs. You might find receivers with aspecification for frequency uncertainty (5 and/or 10 MHz) of about 1 x10-9, even if the specification for their 1 pps output is 100 ns.
Mostreceivers allow the use of a fixed position, after which no furtherpositions are computed. However, some receivers cannot turn offposition fixes, which makes them a poor choice for a frequencystandard. Even though the receiver is stationary, it will appear to bemoving, and the position errors will contribute large fluctuations tothe frequency.
Since each GPS satellite is visible at a givenlocation for a limited time, all GPS receivers must add and removesatellites from the group used to obtain time and frequencyinformation. Often, adding and/or removing a satellite from the timingsolution causes an instantaneous frequency change. Some receivers havemuch better "handoff" algorithms than others.
Differentreceivers handle GPS broadcast errors differently. For example, if asatellite is broadcasting bad data (such as PRN 5 on March 18, 1997),some receivers fail, and others do not. Some receivers have built-insoftware designed to remove "bad" data, but even these receivers mightfail under certain conditions.

If requested by a manufactureror end user, NIST can evaluate a particular GPS receiver for itssuitability as a frequency standard. For a predetermined fee, NIST willissue a report stating the frequency uncertainty of the receiver undertest relative to NIST. The uncertainty will be stated for each outputfrequency over a given measurement interval. The results of thisevaluation are kept confidential and not published by NIST.

作者: occult 時間: 2009-9-10 21:38

一般 GPS 接收器的表現
Since the deactivation of the Selective Availability program on May 2,2000, most GPS receivers now produce a 1 pps output with a standarddeviation of 10 ns or less. Many receivers produce frequency with anuncertainty of < 1 x 10-12 when averaged for one day. Two keyfactors that contribute to receiver performance are the quality of thereceiver's internal oscillator, and the quality of the softwarealgorithms that process data acquired from the satellites.

作者: occult 時間: 2009-9-10 21:38

總結
A properly designed and maintained GPS receiver can be used to showfrequency traceability to NIST at an uncertainty of < 1 x 10-12 whenaveraged for one day. There are at least 2 ways to establishtraceability to NIST through the use of GPS:

By subscribing to the NIST Frequency Measurement and Analysis Service.This service provides everything needed to establish traceability toNIST at an uncertainty of 2 x 10-13, including a complete GPS-basedmeasurement system that can calibrate 5 oscillators at once. Themeasurement results are reviewed by NIST personnel and a monthlytraceability report and uncertainty statement is sent to eachsubscriber.
By using a commercial GPS receiver designed to workas a traceable frequency standard, operating the receiver properly, andestablishing a traceability chain through the use of the NIST GPS data archive,or the use of NIST-USNO comparison data. The uncertainty of a specificreceiver can be estimated by using the manufacturer's specification, orby having NIST or another national measurement laboratory perform anevaluation.

歡迎光臨香港天文學會-掩星組 (http://occultation.freebbs.tw/)