XITRON XT2640 - Synchronizing Measurements between Vpas

XT2640OperatingManual 13July2016 Page115of187

22 SYNCHRONIZINGMEASUREMENTSBETWEENVPAS
TheXT2640canbeconfiguredtosynchronizethestartingtimeofmeasurementsindifferentVPAstoaccommodatemeasurements
ofpowerlossandefficiencywhenthepowerlevelsarevaryingrapidlyovertime.ThisisalwayspossiblewithinasingleXT2640and
optionallyalsobetweenmultipleXT2640susingtheMulti‐Unitcapabilities(seesection21).
ThisisachievedbyselectingtheSYNCmeasurementmodeintheXT2640(forasingleXT2640)orineachXT2640whichitisdesired
tosynchronizewithinaMulti‐Unitgroup.
Note:
NotallunitsinaMulti‐UnitGroupneedtobeconfiguredwiththesamemeasurementmode,thesamemeasurementperiod
orusethesamefundamentalfrequency.
SelectingtheSYNCmeasurementmodeonlyhasanybenefitwhenitisdesiredtoaccuratelytrackpowerlossand/or
efficiencyduringsignificantchangesofpowerlevel.InmostapplicationstheSYNCmodeisnotrecommended.
SynchronizingVPAsinthismannercausesallsuchVPAsexceptfortheonewiththelongestmeasurementperiodtohave
smallgapsbetweentheirmeasurementperiodsduringwhichnomeasurementisbeingmade.Typicallythishaslittle
consequenceandisnotspecifictotheXT2640butappliestoanysynchronizedmeasurementperiodshavingdifferent
lengths.
ThisisnotthesameasselectingforthemeasurementperiodofaVPAtobesynchronoustothatofanotherVPAinthesame
XT2640bysettingthePERIODsettinginoneVPAtoanotherVPAinthesameXT2640(see12.4.2.2.1fordetailsregarding
this)asconfiguringinthatmannerselectsthatbothVPAshaveexactlythesamemeasurementperiodandforcesthemto
useexactlythesamefundamentalfrequencyformeasurementsynchronization(thisisonlyavailablewithinasingle
XT2640).WhenconfiguredforSYNCmode,evenindifferentXT2640s,youcanconfigureanyVPAstousethesame
fundamentalfrequencyandsohavetheexactlythesamemeasurementperiodlengthbyusingtheFUNDconfiguration
settingortheycaneachindependentlyderivetheirownfundamentalfrequenciesasneeded.
WithinasingleXT2640,orwithinmultipleXT2640’sinthesameMulti‐Unitgroup,ifconfiguredinSYNCmeasurementmodeall
VPAsperformsynchronizationcollectivelyasfollows–
AftercompletingeverymeasurementperiodeverysuchVPAchecksifallotherVPAsinthecollectionarepresentlywaiting
aftercompletingtheirmeasurementperiods.IfallsuchVPAshavealreadycompletedtheirmeasurementperiodsthenall
VPAsarecommandedtostartanewmeasurementperiod,otherwisethisVPAwaitstobecommandedbyanotherVPAto
startafurthermeasurementperiodandinformsallotherVPAsassuch(usingamulti‐castpacketwithinaMulti‐Unitgroup
ifsynchronizingmultipleXT2640’s).
InthismannereachVPAmayoptionallybeconfiguredwithadifferentmeasurementperiod(ornot),whetherbysettingadifferent
PERIODsettingforeachorbecauseVPAshavedifferentsignalstowhichtheyaresynchronizingtheirmeasurementperiods(i.e.each
ismeasuringsignalshavingdifferentandunrelatedfrequencycontent).
WithinasingleXT2640thestartingtimesofsuchmeasurementsaresynchronizedtowithin10µs,andforVPAsindifferent
XT2640’susingtheMulti‐UnitcapabilitythisisextendedslightlybytheLANinterfacedelay.Thesynchronizinginformationpassed
betweenmultipleXT2640’sisasingleminimumlengthpacketontheLANinterfacewhichismulticasttoallconfiguredXT2640’s
withthehighestpriority,sotypicallythisincursanadditionaldelayof6.4µstimes(1+thenumberofinterveningswitchesinthe
networktopology).
Theeffectofsuchslightdifferencesbetweenthestartingtimesofthemeasurementsistypicallyquiteminorasdescribedbelow‐
Thesedifferencesonlyeffectmeasurementsofpowerlossandefficiencywhensignalsarechanging;thereisnoerrorifthe
signalsarenotsubstantiallychangingandthereisnoerrorinthemeasurementsofpowerwithineachVPAitselfunderany
circumstances.
SincemultipleXT2640susemulti‐castedcommunicationsforsynchronizationandalmostallswitchesforwardmulti‐casted
packetsvirtuallysimultaneouslybetweenports,thisdelayisnotdependentonthenumberofXT2640sinaMulti‐UnitGroup.
Theworstcasedelay(fora100MLAN)betweenmultipleXT2640’swithasingleinterveningswitchis152µsifbotha)thereis
othertrafficpresentonthatnetworkportionatexactlythesametimeasoneXT2640finishesameasurementperiodandb)
thattraffichasthemaximumpacketlengthandc)isdestinedforoneoftheXT2640sbeingsynchronized.Trafficbetween
othernetworknodeshasnegligibleaffectfortypicalswitchesastheyemploydifferentpacketbuffersandroutingbetween
nodes.
Irrespectiveofthecauseofthesedelayerrors,theyarenotcumulative.I.e.theerrorover5measurementperiodsis1/5
th
of
theerroroveronemeasurementperiod.Inmanycircumstancesitisusefultousethistoadvantagebyselectingafaster
nominalmeasurementperiodineveryVPAandselectingforaveragingusingthesameRESPONSEsettingineachVPA.
IfallsuchVPAsareconfiguredfor100msnominalmeasurementperiodsthentheworst‐caseerrorinpowerlossorefficiency
measurementsis0.001%oftheinter‐measurementchangeperµsofdelaybetweenVPAs(or0.01%ifconfiguredforaperiod
of10ms,or0.00033%ifconfiguredfor300ms).WithinthesameXT2640thisyieldsaworstcaseerrorof0.01%ofthe
change,ortypically0.0164%ofthechangewithdifferentXT2640’swithoutinterveningswitches.Asanexampleusing100ms
measurementperiods,ifapowerlevelsuddenlychangesfrom100Wto200Wduringasingle100msmeasurementperiod,
thenthereisapossibleerrorofupto0.01%of100W(i.e.10mW)inpowerloss/efficiencyinthatone100msmeasurement
periodinwhichthe100Wpowerlevelchangeoccurredbutthereisnoerrorinthemeasuredpowerlossorefficiencyinany
measurementperiodeitherbeforeorafterthechange.
IfsuchVPAsaremeasuringsignalshavingdifferentfrequencycontentthenthedifferencesinmeasurementperiodscausedby
thatdifferentcontentistypicallyofmoresignificancethanerrorscausedbysynchronizationdelays.Thisisbecausethe
measurementperiodsarenotthesameandsoexactsynchronizationisnotpossible.Itisnotpossibletogiveageneraltypical
figureforthiseffect,butthemeasurementperiodforeveryVPAcanbeestimatedbycalculatingtheperiodofthenearest
integernumberoffundamentalcyclesintheconfiguredmeasurementperiodandthencalculatingthedifferencebetweenthe