1INDUSTRIALANDCOLLABORATIVECONTROLSYSTEMS-ACOMPLEMENTARYSYMBIOSIS–-Lookingattoday’scontrolsystemonecanfindawidevarietyofimplementations.Frompureindustrialtocollaborativecontrolsystem(CCS)toolkitstohomegrownsystemsandanyvariationin-between.DecisionsonthetypeofimplementationshouldbedrivenbytechnicalargumentsRealityshowsthatfinancialandsociologicalreasonsformthecompletepicture.Anydecisionhasit’sadvantagesandit’sdrawbacks.Reliability,gooddocumentationandsupportareargumentsforindustrialcontrols.Financialargumentsdrivedecisionstowardscollaborativetools.Keepingthehandsonthesourcecodeandbeingabletosolveproblemsonyourownandfasterthanindustryaretheargumentforhomegrownsolutionsoropensourcesolutions.Theexperienceofmanyyearsofoperationsshowsthatwhichsolutionistheprimaryonedoesnotmatter,therearealwaysareaswhereatleastpartoftheotherimplementationsexist.Asaresultheterogeneoussystemshavetobemaintained.Thesupportfordifferentprotocolsisessential.Thispaperdescribesourexperiencewithindustrialcontrolsystems,PLCcontrolledturnkeysystems,theCCStoolkitEPICSandtheoperabilitybetweenallofthem.-INTRODUCTIONProcesscontrolsingeneralstartedatDESYintheearly80thwiththeinstallationofthecryogeniccontrolsystemfortheacceleratorHERA(Hadron-Elektron-Ring-Anlage).Anewtechnologywasnecessarybecausetheexistinghardwarewasnotcapabletohandlestandardprocesscontrolssignalslike4to20mAinputandoutputsignalsandthesoftwarewasnotdesignedtorunPIDcontrolloopsatastablerepetitionrateof0.1seconds.Inadditionsequenceprogramswerenecessarytoimplementstartupandshutdownproceduresforthecomplexcryogenicprocesseslikecoldboxesandcompetecompressorstreets.Soonitwasnecessarytoaddinterfacestofieldbusesandtoaddcomputingpowertocryogeniccontrols.SincetheinstalledD/3system[1]onlyprovidedandocumentedserialconnectiononamultibusboard,thedecisionwasmadetoimplementaDMAconnectiontoVMEandtoemulatethemultibusboard’sfunctionality.ThenecessarycomputingpowerfortemperatureconversionscamefromaMotorolaMVME167CPUandthefieldbusadaptertotheinhouseSEDACfieldbuswasrunningonanadditionalMVME162.TheoperatingsystemwasVxWorksandtheapplicationwastheEPICStoolkit.SincethisimplementationwassuccessfulitwasalsoimplementedfortheutilitycontrolswhichwerelookingforagenericsolutiontosupervisetheirdistributedPLC’s.2ASELECTIONOFPROCESSCONTROLSYSTEMSATDESYDCS(D/3)AsaresultofamarketsurveytheD/3systemfromGSEwasselectedfortheHERAcryogenicplant.ThedecisionwasfortunatebecauseoftheDCScharacteroftheD/3.Thepossibilitytoexpandthesystemonthedisplay-andontheI/OsidehelpedtosolvetheincreasingcontroldemandsforHERA.ThelimitingfactorforthesizeofthesystemisnotthetotalnumberofI/Obutthetrafficonthecommunicationnetwork.Thistrafficisdeterminedbythetotalamountofarchiveddatanotbythedataconfiguredinthealarmsystem.Thetechnicalbackgroundofthislimitationisthefactthatarchiveddataarepolledfromthedisplayserverswhereasthealarmsarepushedtoconfigureddestinationslikealarm-files,(printer)queuesordisplays.SCADASystemswithDCSFeatures(Cube)ThefactthattheD/3systemmentionedabovehadsomehardcodedlimitationswithrespecttotheY2Kproblemwasforcingustolookforanupgradeorareplacementoftheexistingsystem.AsaresultofacallfortenderthecompanyOrsiwiththeirproductCubecameintoplay[2].Theprojectincludedacompletereplacementoftheinstalledfunctionality.ThisincludedtheD/3aswellastheintegrationoftheDESYfieldbusSEDACandthetemperatureconversioninVME.Theprojectstartedpromising.Butsoontechnicalandorganizationalproblemswerepushingthescheduletoit’slimitswhichweredeterminedbytheHERAshutdownscheduledatthattime.Thefinalacceptancetestatthevendorssiteshoweddramaticperformanceproblems.Twofactorscouldbeidentifiedasthecauseoftheseproblems.ThefirstonewasrelatedtotheunderestimatedCPUloadofthe6thgradepolynomialtemperatureconversionrunningat1Hz.ThesecondonewastheadditionalCPUloadcausedbythecomplexfunctionalityoftheexistingD/3system.Hereitwasunderestimatedthateachdigitalandanaloginputandoutputchannelhadit’sownalarmlimitsintheD/3system.InaSCADAlikesystemasCubethebasefunctionalityofachannelistoreadthevalueandmakeitavailabletothesystem.Anyadditionalfunctionalitymustbeadded.Lastnotleasttheloadonthenetworkforpollingallthealarmlimits–typicallyforaSCADAsystem–wasalsodrivingthenetworktoit’slimits.FinallythecontractwithOrsiwascancelledandanupgradeoftheD/3systemwastheonlypossiblesolution.Itwasfinallycarriedoutinmarch2003.InanycaseitshouldbementionedthattheCubeapproachhadtheadvantageofahomogeneousconfigurationenvironment(fortheCubefrontendcontrollers)–comparedwithheterogeneousenvironmentsfor‘pure’SCADAsystems.SCADA(PVSS-II)TheH1experimentattheHERAacceleratordecidedtousePVSS-IIforanupgradeoftheirslowcontrolsystems[3].TheexistingsystemsweredevelopedbyseveralmembersoftheH1collaborationandweredifficulttomaintain.The3decisiontousePVSSasareplacementwasdrivenbytheresultsofanextensivesurveycarriedoutatCERNbytheJointControlsProject[4].PVSSisa‘pure’SupervisoryAndDataAcquisitionSystem(SCADA).ItprovidesasetofdriversforseveralfieldbusesandgenericsocketlibrariestoimplementcommunicationoverTCP/IP.Thecoreelementisthesocalledevent