Holdover in synchronization applications: Difference between revisions

Revision as of 04:00, 2 June 2022

Latest revision as of 01:30, 5 June 2025

(8 intermediate revisions by 7 users not shown)

Line 1:

{{Use dmy dates|date=September 2022}}

Two independent clocks, once synchronized, will walk away from one another without limit.<ref name="smartclock">{{cite journal|url=http://tf.nist.gov/general/pdf/988.pdf |title=Smart Clock: A New Time |publisher=[[IEEE]] |date=1992-12-06 |bibcode=1992ITIM...41..915W |access-date=2012-10-21 |last1=Weiss |first1=M. A. |last2=Allan |first2=D. W. |last3=Davis |first3=D. D. |last4=Levine |first4=J. |journal= IEEE Transactions on Instrumentation and Measurement|volume=41 |issue=6 |page=915 |doi=10.1109/19.199433 }}</ref> To have them display the same time it would be necessary to re-synchronize them at regular intervals. The period between synchronizations is referred to as '''holdover''' and performance under holdover relies on the quality of the reference oscillator, the PLL design, and the correction mechanisms employed.<ref name="analog1">{{cite web|url=http://www.analog.com/static/imported-files/application_notes/AN-1002.pdf |title=AN-1002 (Rev. 0) |access-date=2012-09-28}}</ref>

==Importance==

Line 17 ⟶ 18:

Within the base station, besides standard functions, accurate timing and the means to maintain it through holdover is vitally important for services such as [[E911]]<ref name="eetimes1"/>

GPS as a source of timing is a key component in not just [[Synchronization in telecommunications]] but to critical infrastructure in general.<ref>{{cite journal|url=https://apps.dtic.mil/sti/pdfs/ADA503921.pdf |archive-url=https://web.archive.org/web/20121009132113/http://www.dtic.mil/cgi-bin/GetTRDoc?AD=ADA503921 |url-status=live |archive-date=October 9, 2012 |title=Global Positioning System Timing Criticality Assessment - Preliminary Performance Results |author1=Dr. James Carroll |author2=Mr. Kirk Montgomery |access-date=2012-10-21}}</ref> Of the 18 Critical Resource and Key infrastructure (CIKR<ref>{{cite web |url=http://training.fema.gov/EMIWeb/IS/IS860a/CIKR/sectorMenu.htm |title=ArchivedCIKR Resource copyCenter |access-date=2011-09-12 |url-status=dead |archive-url=https://web.archive.org/web/20111016092905/http://training.fema.gov/EMIWeb/IS/IS860a/CIKR/sectorMenu.htm |archive-date=2011-10-16 }}</ref>)sectors, 15 use GPS derived timing to function correctly.<ref name="autogenerated2">{{cite web |url=http://www.swpc.noaa.gov/sww/SWW_2011_Presentations/Wed_830/GPS-PNTTimingStudy-SpaceWeather4-27.pptx |title=GPS Critical Infrastructure |author=R. James Caverly |date=2011-04-27 |access-date=2012-10-21 |archive-url=https://web.archive.org/web/20120324093146/http://www.swpc.noaa.gov/sww/SWW_2011_Presentations/Wed_830/GPS-PNTTimingStudy-SpaceWeather4-27.pptx |archive-date=2012-03-24 |url-status=dead }}</ref> One notable application where highly accurate timing accuracy (and the means to maintain it through holdover) is of importance is in the use of [[Synchrophasor]]s in the power industry to detect line faults.<ref>{{cite web|url=http://tf.nist.gov/general/pdf/2193.pdf |title=Signal |access-date=2012-09-28}}</ref> Another is in [[Low latency (capital markets)|Low latency]] trading applications in capital markets.

==How GPS-derived timing can fail==

Line 52 ⟶ 53:

[[NIST]] defines a Disciplined Oscillator as:

<blockquote>

''An oscillator whose output frequency is continuously steered (often through the use of a [[phase locked loop]]) to agree with an external reference. For example, a GPS disciplined oscillator (GPSDO) usually consists of a quartz or rubidium oscillator whose output frequency is continuously steered to agree with signals broadcast by the GPS satellites.''<ref>{{cite web |url=http://tf.nist.gov/general/enc-d.htm |title=Time and Frequency from A to Z |publisher=Tf.nist.gov |access-date=2012-09-28 |archive-url=https://web.archive.org/web/20120911011249/http://tf.nist.gov/general/enc-d.htm |archive-date=2012-09-11 |url-status=dead }}</ref>

</blockquote>

Line 59 ⟶ 60:

[[File:GPSDO.png|thumb|left|A Modern [[GPSDO]]]]

Amongst the building blocks of a GPS Time and Frequency solution the oscillator is a key component<ref name="autogenerated2"/> and typically they are built around an Oven Controlled Crystal Oscillator ([[Crystal oven|OCXO]]) or a [[Rubidium standard|Rubidium based clock]]. The dominant factors influencing the quality of the reference oscillator are taken to be aging and temperature stability. However, depending upon the construction of the oscillator, barometric pressure and relative humidity can have at least as strong an influence on the stability of the quartz oscillator.{{citation needed|date=December 2012}} What is often referred to as "random walk" instability is actually a deterministic effect of environmental parameters. These can be measured and modeled to vastly improve the performance of quartz oscillators. An addition of a Microprocessor to the reference oscillator can improve temperature stability and aging performance<ref>{{cite book|doi=10.1109/FREQ.2008.4622980 |isbn=978-1-4244-1794-0 |publisher=Ieeexplore.ieee.org[[IEEE]] |chapter=Improvements in OCXO performance by the use of an on-board microprocessor |title=2008 IEEE International Frequency Control Symposium |year=2008 |last1=Wacker |first1=Mike F. |last2=Villella |first2=A. |pages=159–164 |s2cid=46081633 }}</ref> During Holdover any remaining clock error caused by aging and temperature instability can be corrected by control mechanisms.<ref>{{cite journal|url=http://kunz-pc.sce.carleton.ca/thesis/CrystalOscillators.pdf |title=Frequency Accuracy & Stability Dependencies of Crystal Oscillators |author1=Hui Zhou |author2=Charles Nicholls |author3=Thomas Kunz |author4=Howard Schwartz |date=November 2008 |access-date=2012-10-21}}</ref> A combination of quartz based reference oscillator (such as an [[Crystal oven|OCXO]]) and modern correction algorithms can get good results in Holdover applications.<ref name="Lombardi">{{cite journal|url=http://tf.nist.gov/general/pdf/2297.pdf |title=The Use of GPS Disciplined Oscillators as Primary Frequency Standards for Calibration and Metrology Laboratories |author=Michael A. Lombardi |publisher=NCSL International |date=September 2008 |access-date=2012-10-21}}</ref>

The holdover capability then is provided either by a free running local oscillator, or a local oscillator that is steered with software that retains knowledge of its past performance.<ref name="Lombardi"/> The earliest documentation of such an effort comes from the then National Bureau of Standards in 1968 [Allan, Fey, Machlan and Barnes, "An Ultra Precise Time Synchronization System Designed By Computer Simulation", Frequency], where an analog computer consisting of ball-disk integrators implemented a third order control loop to correct for the frequency ageing of an oscillator. The first microprocessor implementation of this concept occurred in 1983 [Bourke, Penrod, "An Analysis of a Microprocessor Controlled Disciplined Frequency Standard", Frequency Control Symposium] where Loran-C broadcasts were used to discipline very high quality quartz oscillators as a [[caesium]] replacement in telecommunications wireline network synchronization. The basic aim of a steering mechanism is to improve the stability of a clock or oscillator while minimizing the number of times it needs calibration.<ref name="smartclock"/> In Holdover the learned behaviour of the [[Crystal oven|OCXO]] is used to anticipate and correct for future behavior.<ref name="analog1"/> Effective aging and temperature compensation can be provided by such a mechanism<ref>{{cite journal |url=http://www.gmat.unsw.edu.au/snap/publications/tappero_etal2007c.pdf |author1=Fabrizio Tappero |author2=Andrew G. Dempster |author3=Toshiaki Iwata |title=Phase error reduction method for free-run QZSS clock |publisher=[[IEEE]] |date=2007 |access-date=2012-10-21 |url-status=dead |archive-url=https://web.archive.org/web/20130512052335/http://www.gmat.unsw.edu.au/snap/publications/tappero_etal2007c.pdf |archive-date=2013-05-12 }}</ref> and the system designer is faced with a range of choices for algorithms and techniques to do this correction including extrapolation, interpolation and predictive filters (including [[Kalman filter]]s).<ref>[http://www.eftf.org/proceedings/PDFs/FPE-0031.pdf Proceedings] {{dead link|date=September 2012}}</ref><ref>{{cite book|doi=10.1109/FREQ.2004.1418510 |isbn=0-7803-8414-8 |publisher=Ieeexplore.ieee.org[[IEEE]] |chapter=Adaptive OCXO drift correction algorithm |title=Proceedings of the 2004 IEEE International Frequency Control Symposium and Exposition, 2004 |year=2004 |last1=c.w.t. Nicholls |first1=G.C. Carleton |pages=509–517 |s2cid=33745348 }}</ref>

Once the barriers of aging and environmental effects are removed the only theoretical limitation to holdover performance in such a GPSDO is irregularity or noise in the drift rate, which is quantified using a metric like [[Allan deviation]] or [[Time deviation]].<ref>[http://www.leapsecond.com/pages/adev/adev-why.htm Leap second]</ref>{{Unreliable source?|date=October 2012}}

The complexity in trying to predict the effects on Holdover due to systematic effects like aging and temperature stability and stochastic influences like [[Random walk|Random Walk]] noise has resulted in tailor-made ''Holdover Oscillator'' solutions being introduced in the market.<ref>{{cite web|url=http://www.vectron.com/products/modules/MD-023.pdf |title=MD-023: Extended Holdover Crystal Oscillator |publisher=Vectron International |date=June 2011 |access-date=2012-10-21}}</ref>