Fiber Tracker Meeting Minutes

June 24, 2003 - 10am, Farside

• Data quality, on-line view : Warchol
• talk slides
• want fast detection of hardware problems
• stress importance of watching pdaq_examine - VRBC gets out-of-sync and reads wrong event
• Tom Diehl: Is there a way to easily identify when wrong events are read out?
  • Drew: It's in the raw data, but there is no advertized way to get at it offline right now
• Might have found a cure for the intermittant occurance of VRBC sync problems - info in an email from Dave?
• What remains to be done to improve online data quality?
• optimization of bias voltage
• optimization of threshold settings (get rid of infamous "20 ADC cut")
• improve discriminator quality (using a threshold curve? no concept at present)
• do optimization as a function of luminosity
• Alan: Any news on pedistal problem investigation?
• Fred: Not much has happened in last four weeks; however, offline tracking efficiency is high so it's not hurting us that much at the moment. The extra channel readout problem is getting higher priority right now.
• Marj: Who is working on the alarms?
• Geoff Savage and Jadzia, but they can't claim full responsibility. Needs more people
• Fred: need to switch some focus from making sure online data is good, to how to let people offline know if data is good.
• Runs to Quality : Tom
• talk slides
• need to label data as: bad, reasonable, as good as it gets,...
• Mike Hildreth offers an offline perspective:
• want to know for each detector if there were (a) no hardware problems, or (b) some hardware problems, or (c) catastrophic hardware problems
• there are two databases
• "run query" is filled by software
• "run quality" is filled by hand by people
• all runs (not just physics runs) are available for a run quality search, it's just that most come back as status "UNKNOWN"
• Mike Hildreth: Is there a plan to do statistical test of the histograms?
• Tom: Offline shifters do recocert, but is this enough?
• Mike: Suggest that if such a tool is developed, it be global and not separately written for each system
• General Discussion
• We have software for counting dead and hot wires by David Lam
• It is not ready for production yet (but cft part is)
• steps to reach production:
• do something with the fps
• cps parts need tweaking
• add ability to write stuff to database?
• Drew's suggestions for criteria for hot and dead channels:
• Good = <1% dead or hot (one bad AFE board will make a run less than good, one bad SVX won't)
• Reasonable = <10% dead or hot
• Bad = >10% dead or hot
• Stefan's CTT suggestion:
• Good = all boards are present, and crate x13 is in
• Reasonable = crate x13 missing, but all boards present
• Bad = any board missing
• CTT Db Interface : Stefan
• CTT has an expert translating the checklists and logbook entries into database entries about once a week
• instruction page about this: http://www-d0online.fnal.gov/www/groups/cft/CTT/online/Experts/Tools/DbUpdate/DbUpdate.html
• CTT Analyze : Yurii
• example plots produced by CTT Analyze
• relies on tracking algorithm and crate x13 info
• plan to produce a text table of efficiencies for each TTK term, tracking efficiencies, etc.
• Package information: http://www-do.fnal.gov/~maravin/ctt_analyze/index.html
• More general discussion
• Fred and Drew (and Jadzia?) will help Tom write up what needs to be done
• Drew: Do we want to add some version of what Shabnam has been running to recocert? yes
• Work for Michael and Sangjoon(sp?): Run David's program as it is on current runs (working backwards) and make entries in quality database ASAP
• Need to pick point to break between using Drew's notes and using Stefan's coarse method
• Need to get Brigitte's gui working for CFT input into the database

---

Previous meeting

Index

Next meeting