mTCAuse in Linac to replace CAMAC
Tom HimelJune 4, 2012
1
Outline
General plan for replacing the linac CAMAC.Can it be done without disrupting the accelerator run schedule?Budget and schedule
2
General plan (1 of 3)
As mentioned in my introductory talk:Had planned to replace CAMAC in sectors 10-20 during four 7 month long annual downtimes of FACET.This would save cost of doing the phase I upgrade for LCLS-IIWould allow LCLS-II to start with new hardware.Very unlikely to happen on this time-scale as all available resources are needed for upgrades more essential for LCLS-II
3
General plan (2 of 3)
Full CAMAC replacement is now far enough in the future, have not made new detailed plans.Do want intermediate improvements to be compatible with future plansE.g. LCLS-II linac needs BPMs with better resolution and lower charge sensitivity than present CAMAC readout.Best if done with final architecture so infrastructure can be shared with other systems and have best most modern hardware
4
General plan (3 of 3)
If lab decides to use sectors 1-10 for FACET-II followed by LCLS-III, will propose to do those 10 sectors first during the long downtimes of FACET, then spread to rest of linac after we have all procedures down pat.Otherwise will start in LCLS I or II sectors during down days and the 2 month annual downtimes.Systems where we are happy with LCLS-I solution will mostly be duplicated. Systems with no or problematic LCLS-I solutions will be redone, typically inmTCAHigh channel count systems are the most important to develop early.
5
Plans by system
LLRF:LCLS-I still depends on CAMAC. PADs and PACs supplement the CAMAC on a few stations. All interlocks, timing and some analog readings are done with PIOP and MKSUReplace withmTCAand MKSU-II (interlock and diagnostic network attached device (NAD) developed as part of the LLRF AIP project)BPM:Upgraded linac LCLS-I has VME + NAD which works well. Cannot do 2 bunch, verykludgey.Replace CAMAC withmTCA.Present CAMAC system inadequate for low current running.
6
Plans by system
Quads/Bends:Upgraded linac LCLS uses NAD which works wellDuplicatePresent CAMAC + rack mounted controller solution has regulation problems at low current.CorrectorsUpgraded linac LCLS uses VME ADC and DAC and PLC to control 2 bulk supplies and 16 corrector channels. Works fine but expensiveAIP developing new controller that goes in the crate with the 16 channels and replaces the VME and PLC functionality. Makes the power supply crate a NAD similar to that for the Quad power supply controllerWill use the AIP solution
7
Plans by system
Machine ProtectionUpgraded linac LCLS uses home-brew “link node” chassis that has plug in cards some of which are COTS, others are home-brew. Works well.Will probably duplicate as requirements are odd enough that COTS solution is unlikelyVacuum:LCLS non-linac uses signal conditioning chassis + COTS PLC’sduplicate
8
Plans by system
Temperatures and other slow ADCsLCLS non-linac usesBeckoffI/O. Cheap, works well.DuplicateDigital I/OLCLS non-linac uses IP cards on VME carrier boardWill use same IP cards onmTCAcarrier.
9
Plans by system
Small channel count diagnostics liketoroids, Faraday cups, profile monitors, wire scannersLCLS solutions have various problemsWill improve for LCLS-II as allowed.Fortoroidwould build a new RTM for same ADC used for LLRFLinac upgrade will either duplicate the improved versions or, if needed, develop improved versions probably inmTCA.
10
Installation without disruption
We have considerable experience at this.Process:Test in labTest in part of linac not in useWhere possible (not too many cables), develop quick switch over from old to new and backInstall and test on down daysPhase I was done this way with virtually no disruption of accelerator program. See “Changing horses mid-stream” athttp://www.esrf.eu/icalepcs2011/papers/tucaust04.pdf
11
Installation without disruption
Most RF stations can be taken offline, worked on and put back online while the linac is running. Main exceptions are injector klystrons.ThemTCALLRF testing was done this way using cable adaptors to make the switching of cables quick.Some systems like vacuum and MPS will have to have their final connections made during the two month downtimes.
12
Costcomparision
The two biggest (and hence most expensive) systems are LLRF and BPMs.They are also the only systems for which we have seriously looked at more than one solution.Dan andQiaoshowed that the costs for the considered solutions were quite similarExpect the same to be true for other locations of the same types of hardware
13
Linac upgrade cost and schedule
As I mentioned before, we have not updated the cost and schedule for upgrading the linac as the start date is distant and uncertain.I’ll show some of what we presented at the last review.Mention a few possible changes
14
Budget Cost Analysis
Resource Cost EstimatesAll tasks detailedResources identifiedDiscussed details w/ resources for estimatesSummed resources vs. draft schedulePerformed similar exercise for RF (already approved AIP)Totaled all manpower costs vs. time (24 mo.)
Hardware Cost Estimates
MTCA InfrastructureCOTS based on actual quotes or vendor budgetary quotesIncludes crate, power modules, Hub Controller, Processor IOC, Timing ModuleTiming module not yet a commercial item but base cost on similar commercial products
Hardware Costs 2
Applications ModulesAMCs: ADC-DAC, 3-IP, PMC adapterADC-DAC: Vendor quotes, DESY target (Strück); Vadatech quote pending3-IP: quote in hand (TEWS)PMC: Quotes pending (Vadatech,TEWS)US 2ndsources:Buying first articles to specification, compatible RTMTimingSource is Stockholm U., working to develop vendors
Estimated 10-Sector CostT. Himel
Assume LI11-20 Column costsSolid state sub-boosterrequiredModulator PLCnot requiredBCS & Power Supplies IncludedRF Station Unit Cost = $28K
1440
1440
13110
Summary
The R&D for the major systems is well advanced.We still have a few implementation decisions to make on the minor systemsWe have a rough and somewhat out of date cost estimateWe have experience doing upgrades during short downtimes and while this one involves more cables, it is still doable
19
0
Embed
Upload